Sie sind auf Seite 1von 310

LIBRARY

VAt

POSTGRADUATE

MONTEREY, CA!

ft^aifffnkL
t.fw^'
IS
CONFroENTIAL
A THEORETICAL STUDY OF AUTOMATIC INERTIAL NAVIGATION
by
BJ3., United States Naval
Jr,

Leonard Erb
Lewis
J.

Academy, 1941
Academy, 1941

Stecher,

BJS., United States Naval

SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE
at

MASSACHUSETTS INSTITUTE OF TECHNOLOGY


1949

UMMeMM

This document contains information affecting the national defense of the Unitea States within the meaning of the Espionage Act 50 U.S.C., 31 and 32 as amended. Its transmission or the revelation of its contents in any manner to an unauthorized person is prohibited by law.

*fi&tm\Ai

May

20, 1949

Prof. Joseph S. Newell

Secretary of the Faculty

Massachusetts Institute

of

Technology

Cambridge

39,

Massachusetts

Dear Professor Newell:


accordance with the regulations of the faculty, we hereby submit a thesis entitled, A THEORETICAL
In

STUDY OF AUTOMATIC INERTIAL NAVIGATION


tial fulfillment of the

in

par-

requirements for the degree of

Master

of Science.

tgijteslFffiD

ACKNOWLEDGMENT

The authors wish


to Dr. C. S.

to
to

express their deep appreciation


Dr. W. Wrigley for the interest,

Draper and

confidence, and assistance given by them in supervising the

progress of this work.


Invaluable aid

was also given by

the staff of the

Instrumentation Laboratory, particularly of the "Febe"


Section.
Individual mention is foregone only because of the

many members who


The

helped.

staff of the Rockefeller

Analyzer, at the

Massachusetts

Institute of

Technology, by their interest and

cooperation, contributed greatly to the work.

Thanks are given


ates of Jackson
of the text

to

Mr. L. E. Payne and

his associ-

& Mo re land

for their part in the preparation

and illustrations.

iv

awossffiSR
ABSTRACT
The purpose
of this thesis

was

to postulate

a method of long-range

inertial navigation, using only acceleration inputs,

and

to

examine theo-

retically the response of this

system

to external disturbances, using vari-

ous types

of

mechanizations. The problem was divided into one of track

control and of range indication, using a great-circular path between the


points of departure and destination.

The response

of the

track control system was examined with mecha-

nization equations of as high an order as the fifth, and

was found

to
to

improve
impulse

as the order of the mechanization equation increased. Response


type wind acceleration disturbances

was

satisfactory, but long-period sinuIt

soidally varying winds caused excessive errors.

appears probable that

additional feedback loops, in conjunction with a mechanization equation of


the fifth or higher order, will solve the track control
torily.

problem satisfac-

In the

range indication system,

it

was found

to be

impossible to re-

move any forcing


change

function terms of higher order than the time rate of


It

of acceleration.

appears probable that the range indication sysif

tem, as postulated, will prove to be satisfactory, especially


tivities

the sensi-

are made variable, as a function

of the angle of the input

pendulum.

CONFIDENTIAL

TABLE OF CONTENTS

Chapter
I

Title

Author
Joint
Joint

Page
1

INTRODUCTION
KINEMATICS
1.

17

Purpose
Kinematic Equations
of

2.

Motion Per-

pendicular to Great Circle Track


3.

Longitudinal Motion Kinematic Equations

THE TRACK CONTROL PROBLEM


1.

Erb

31

Introduction

2.
3.

Proportional Control
Integral Control

4.
5.

Second Integral Control


Derivative Control

6. 7.

Proportional Plus Integral Control Proportional Plus First and Second


Integral Control

8.

Proportional Plus First, Second and

Third Integral Control


9.

Track Control Correction as a Fun of Wind Track Control Closed Loop System

tion

10.

11.

Numerical Solutions, Using Two Quadratics


vi

Chapter
12.

Title

Author

Page

Numerical Solutions, Using Four


First Order

Terms

13.

Numerical Solutions, Using Quadratic and Two First Order Terms


Plots of Results

14.

IV

THE RANGE INDICATION PROBLEM


1

Stecher

67

Introduction

2.
3.

Trial Mechanizations

Implementation

of

Equations

4.
5.

The Closed Loop System


Selection of Numerical Values

CONCLUSIONS AND SUGGESTIONS


1.

Joint

103

Disturbances to the Control System


Suggestions for Improvement of

2.

System Response
3.
4.

Track Control Mechanization Summary Range Indication Mechanization Sum-

mary

Appendix

Title

Author
122
Joint

DISCUSSION OF SIMPLIFYING ASSUMPTIONS


1.

Omission

of the

Geodesic Acceleration

Term
2.

Compensation for Coriolis Acceleration


vii

Appendix
3. 4.
5.

Title

Author

Page

Small Angle Pendulum Assumption


Neglect of Vertical Acceleration

Assumption
Response

of Perfect

Aerodynamic

6.

Assumption of Perfect Gyros and Accelerometers

7.

Omission

of Constants of Integration

BIBLIOGRAPHY

136

GLOSSARY OF SYMBOLS AND TERMS

138

viii

CHAPTER

INTRODUCTION

The importance

of developing a

long-range self-contained system of

automatic, mid-course, navigation for the control of guided missiles cannot be overestimated (Fig 1*1).

As long range

flights of piloted aircraft

over enemy territory become increasingly hazardous, because of the necessity of travelling without fighter protection

toward heavily defended cities

provided with elaborate warning nets, automatic navigation can provide a

method
pilots.

of delivering

bombs

without risking the lives of highly trained

With

the

development

of long-range jet-propelled missiles, automatic

navigation should eventually permit the United States to


the

bomb any

place in

world from bases located within

its territorial limits.

For such a system

of automatic navigation to be successful, it

must
It

have a degree of accuracy comparable to that achieved by human pilots.

must not be easily susceptible

to

"jamming" by enemy action.

It

must op-

erate successfully at supersonic velocities, and in the upper reaches of the


earth's atmosphere.

At present, the problems

of

automatic long-range navigation are under

consideration by several groups.


1.

Among

these are:

Baird Associates, Inc.

2.

Hughes Aircraft Co.


Instrumentation Laboratory, M.I.T.

3.

4.
5.

Kollsman Instrument Company


North American Aviation, Inc.

6.

Northrup Aircraft,

Inc.

NHIitNIIAh
7.
8.

Rand Corporation
and Development Sub -office (rocket), Fort Bliss, Texas
U. S.

Army Research

Little

work, however, appears to have been done on the derivation of

theoretically
this

optimum control equations or operating parameters. For


to

reason the writers determined

examine a long-range automatic

navigation system now under development at the Instrumentation Laboratory, M.I.T., as a

means

for exploring theoretically

some

of the general

problems
ect,

of automatic navigation.
is

The Instrumentation Laboratory proj-

which

being executed under

USAF

Contract W33-038ac-13969,
-

is

designated "An Automatic Navigation System


the subject of a report soon to be published.

Project Febe," and will be


its

This project has for

pri-

mary purposes
1.

to

determine:

the feasibility of long-range automatic navigational guidance of bomber airplanes

2.

useful design parameters for a serviceable for military use.

system

Considerations other than military and tactical indicated that the sys-

tem should employ solar tracking, a magnetic azimuth system, and a constant ground speed.

The system

is

allowed to make no contact with the

earth. Altitude is determined by the use of a barometric altimeter; the

only inputs to the system are celestial observation, the magnetic field of
the earth, and the various accelerations experienced

by

the airplane.

This system, which has been installed in an Air Force B-29 bomber,
causes the airplane
to fly

a definitely programmed great circle course at

constant ground speed. Vertical accelerations are

assumed

to

be of such

small importance that they can be neglected.

Figure 1-2 shows a functional

diagram

of the

Febe automatic navigational system. During the period from

CONFIDEWiAi.

z: o LU

o > < U_ oz o X <|< 2 2o h1 _J <r <


cvicr
u.
,

CQ LU U_

zq: oo u_
h^i UJ
>-

o ?:
00

3H b_ CO

CONFIDENTIAL
June to August of 1948, the writers of this thesis

were concerned with

the

Febe project

in

an under -instruction status. Through the courtesy of Dr.

C. S. Draper, they were also permitted to attend a Seminar on Automatic


(Celestial and Inertial)

Long-Range Guidance Systems conducted at

the

Massachusetts

Institute of
1

Technology by the Scientific Advisory Board


3, 1949.

from February

to

February

This background naturally led the writers


the

to

concern themselves with


take.

form which a practicable service guidance system might

The

Febe system presents several technical and military problems, which


arise principally in connection with the following:
1.

Constant ground speed

2. 3.

The necessity

of solar tracking

Magnetic azimuth input

4.

Weight and size

of

components.

Since a successful automatic navigational system is ultimately destined to fly in a long-range high-speed missile,
the
it

must be assumed

that

weapon

will be operating at close to the

maximum

range permitted by

its size

and fuel capacity. The navigational system, therefore, should not


This suggests a system which will
fly

seriously lower the fuel economy.

at constant, or nearly constant, airspeed.

Furthermore,

if

the

power plant

consists of an athodyd, the airspeed must be maintained very nearly constant by the fundamental limitations of this type of propulsion.

The limitation

of the

Febe system

to a constant

ground speed was

largely dictated by the requirements of the azimuth system.

The magnetic

input could be eliminated through the use of two star-tracking telescopes.

The problems

of celestial tracking
is

become increasingly severe, however,


it is

as missile speed

increased. At supersonic and near-sonic speeds,


5'

CONFIDENTIAL
no longer feasible
unit, projecting

to

have an astrodome,
the fuselage.

in

which

to

house the tracking


flat

from

Thus, tracking through a

window,

with the attendant difficulties which this entails,

becomes a necessity.

Also, at high speeds, thick boundary layers and intense heating will exist

along the surface of the missile, greatly complicating optical problems.


Finally, the inclusion of a celestial tracking

system materially increases

the space and weight required by the navigational

system

and weight and

space factors become increasingly important with increasing range and

bomb

load of an aircraft.

It

appears, however, that approximately a thirty -

fold decrease in the uncertainty levels of existing gyros

would permit the

maintenance

of

an inertial coordinate reference system within the missile

through the use of such gyros alone, thus eliminating the use of celestial
tracking.

Great progress

is

being

made

in the

improvement

of gyros.

The work,

for example, that is being done by the group at the Massachusetts Institute
of

Technology under Dr. C.

S.

Draper shows promise

of obtaining the

nec-

essary accuracy within the next few years.

The model
stalled in a
It

of the
is

Febe automatic navigational system presently


lbs.

in-

B-29

very bulky, and weights 1917

(See Fig 1-3).


of the airplane.

practically fills the after pressurized


it

compartment

Of

course,

must be remembered

that no effort

was made

to

decrease the

size of this equipment, or to keep its weight at a

minimum. Nevertheless,

a missile held to a constant airspeed, rather than to a

programmed ground

spped, could, by using gimbal solvers, greatly simplify the elaborate trigo-

nometric and speed computers which Febe requires. Much


the computation

of the

work

of

systems might,

of

course, be performed on the ground

HAL
O

<

=>

LU CD

CO

LU

LU co

>

CO 1
I

< a.

o o Q LU M
Lt Z> CO CO
LxJ

or a.

<
CD OJ
I

o
q:

GO

o h2 O H o o < 2 o o
(A)

2 LU
_l LU

CONFIDENTIAL
prior to the start of the flight, and be fed to the system through tapes, but
this

would introduce further complications.

From
tional

all of these

considerations, a self-contained inertial -gravita-

system

flying a great circle course at nearly constant airspeed

appears

to provide one of the

most satisfactory solutions,

if

such a system

can be implemented.
After this examination of the Febe system, there remain, then, for

use in the purely inertial system, a gyro-stabilized inertial platform, a


clock

mechanism

to

remove

the earth's diurnal rotation, and two mutually

perpendicular single -degree -of -freedom pendulous accelerometer units.


In addition, there are

gimbals and servomechanisms isolating the inertial


it,

platform from the motions of the missile containing

and orienting the

platform so that

its

axis is parallel to the polar axis of the earth, so that

a reference vertical parallel to that of the point of departure (or any other

convenient reference direction) is maintained independent of the motion of


the missile.
(See Figs
1-4, 1-5

and 1-6.)
to maintain a

The three integrating gyros which are used


mounted inertial platform "fixed"

gimbal-

in inertial space, parallel to the earth's

equatorial plane, are mounted so that they detect motion of the platform

with respect to inertial space about the polar axis of the earth, and about

two mutually perpendicular axes parallel


1-7, 1-8).

to the plane of the

equator (Figs.
the

Servomechanisms, controlled by these gyros, rotate

gimbals

so that the inertial platform remains fixed in inertial space. The earth
platform, mounted on the inertial platform with the identical polar axis, is
rotated about that axis once in twenty -four hours, to transform the inertial

reference system of the inertial platform into an earth reference system.

\
TRACK CONTROL PENDULUM UNIT

RANGE INDICATION

PENDULUM UNIT

CONTROLLED MEMBER

CONTROLLED MEMBER
GIMBAL SYSTEM BASE

CONTROLLED MEMBER OUTER AXIS GIMBAL

INNER

ISOLATION
G

GIMBAL

/
1
1

12

INCHES

FIG.

1-4.

PHOTOGRAPH OF INERTIAL REFERENCE AUTOMATIC NAVIGATION SYSTEM MODEL.

CO
QR)(tcpl

TRACK CONTROL

SYSTEM
icmlitc)

(S9)(DC) (ind)ltcU

PENDULOU' UNIT

TRACK CONTROL MECHANIZATION

(5<))Az(cm)lind)

COMPARATOR

ISqJClA Z(cm)

ROLL AND
DRIVE

YAW SYSTEM

X'lcm|

5<fAz(cm)(true)
COMPARATOR

CD
SEE GLOSSARY (APPENDIX i_) FOR DEFINITION OF EXCEPT FOR THOSE LISTED BELOW.
"INERTlAL REACTION OF GRAVITY
IN

TERMS

GIMBAL

CONTROLLED

(cm)

SYSTEM
(FOR CONTROLLE

lEPlles)

MEMBER

EARTH PLATFORM

JlPlllS)

GIMBAL
E

AND SYMBOLS
M(is)

COMBIMOR

UPlUsi

INERTlAL

(DIP,(IP)

(is)

ISOLATION

PLATFORM
=P
1

SYSTEM
(FOR INERTlAL

MISSILE (WITH

_^6rt)(trueHtc)

Q(IR)Ucp)

TRACK

CONTROL
SURFACES)

CONTROL PLANEVECTOR POSITION


(l>\J(dep) "POLAR
'

MEMBER

(JPKIS)

-P M(LS)

PLATFORM)
I

IN

ONE PLANE.

(Sg)(PAt
EARTH POLAR

ANGLE AT POINT OF OEPARTURE.

(Sq)iCD)
PAlfclep)

_ dndl
tinqi

AXIS

CONTROL

DIRECTION ANOLE RESOLVER

TIME

GYRO SYSTEM

(OP

Az(GCXjJep)" AZ "v1u rH 0F ANGLE OF THE GREAT CIRCLE TRACK AT THE POINT OF DEPARTURE
(1

Pill SI

>E(set)

RADIUS OF EARTH SETTING.

(Sq)
^Z(GCHdep)

-DENOTES A SIGNAL
-DENOTES INERTlAL SPACE
-

A
cosA [BW t _-JCD|ji ILd|L.
CORIOLIS

r (indl

(is)

(es)

DENOTES EARTH SPACE


OF EARTH WITH RESPECT TO INERTlAL SPACE.

1E

-ANGULAR RATE OF ROTATION


-OEVIATION

'Elsetl

ACCELERATION CORRECTION

(Sq)A rlmdl

(D)

SYSTEM

(0

-CORRECTION

RANGE
(cm|(ri|

INPUTS

INDICATION

(Sql(DC)

QlIR) (npl

SYSTEM PENDULOUS
UNIT

(indlllonq^

LONGITUDINAL MECHANIZATION

(Sq)Ar(md)
_

RANGE ANGLE DRIVE

^-

SIGNAL OUTPUTS

SYSTEM

^Mjyert)
jdestj - (V e
r

PHYSICAL CONNECTION

^~ POWER
r

LEVEL OUTPUTS

Ildestl-ldepl

COMPARATOR

Uind][

MEASURE

OF

DISTANCE TO DESTINATION

FIG.I-5.

SIMPLIFIED

FUNCTIONAL DIAGRAM OF AUTOMATIC INERTlAL GUIDANCE SYSTEM.


10

CONFIDENTIAL
EARTH POLAR AXIS
GREAT CIRCLE COURSE BETWEEN DEPARTURE ANO DESTINATION POINTS
DESTINATION POINT
CONTROLLED

(PA) (dep)

MEMBER

DEPARTURE
POINT
d P> NCC VERTICAL)
(

/OUTER

AXIS OF GlMBAL

OP CONTROLLED

MEMBER

FIGURE 1-6

GEOMETRIC DIAGRAM OF AXES OF AUTOMATIC NAVIGATION SYSTEM


u

SINGLE -DEGREE -OF FREEDOM GYRO UNITS

NERTIAL PLATFORM

OUTER
ISOLAT ION GIMBAL

BASE

FIG. 1-7.

PHOTOGRAPHIC VIEW OF INERTIAL REFERENCE AUTOMATIC NAVIGATION SYSTEM MODEL WITH POLAR AXIS TIPPED AWAY FROM THE NORMAL DIRECTION TO THE BASE.

ISOLATION CONTROLLED MEMBER

GIMBAL

BASE
/

-;
1 1

12

INCHES

FIG.

1-8.

PHOTOGRAPHIC TOP VIEW OF INERTIAL REFERENCE AUTOMATIC NAVIGATION SYSTEM MODEL WITH CONTROLLED MEMBER GIMBAL SYSTEM REMOVED.

CGNFJDEN HAL
Upon
in
the earth platform, in turn, the controlled
1-9).

member

is

mounted
is

gimbals (Fig

The axis

of the outer

gimbal of this

member

made

to lie in the plane of the great circle connecting the points of

de-

parture and destination on the surface of the earth, considered as a

sphere.*

Then, about the inner controlled-member gimbal axis, the con-

trolled -member is

moved

until it lies parallel to the

surface of the earth


is

at the point of departure.

The inner controlled-member axis

placed

perpendicular to the

programmed great

circle path so that, as the missile

moves along

the

programmed

path, the controlled

member, by

rotation

about the inner axis alone, can be maintained at all times parallel to the
surface of the earth below the missile.
trolled

The angle through which

the con-

member has

rotated will then be a direct reading of the distance

travelled by the missile along the

programmed great

circle.
the

At some pre-

determined angle between the controlled


ing
it

member and
may be

gimbal connect-

to the earth platform, the missile

said to have arrived at its

destination.

The two pendulous units are mounted upon the controlled member, one
with
its

plane of motion in that of the

programmed great

circle, the other

with

its

plane of motion perpendicular to that great circle plane.

The lon-

gitudinal pendulum, i.e. the unit with its input plane parallel to the plane
of the

programmed

track, sensing longitudinal accelerations, produces an

output signal which is then modified through the implementation of the longitudinal mechanisation equation.

The resultant signal

is

used as the input

This orientation is accomplished by rotating the controlled member on the earth platform to the correct meridian angle of the great circle track at the point of departure, LA and by mtin % e outer gimbal z(o,c.)(dep)^ axis to the correct polar angle (PA)/. on v of the great circle track plane with respect to the polar axis of the *aep; earth.
*

UJ

o <

to

CO ID

O
_l

Q
LU

UJ

Q O
UJ CD

UJ

Q
UJ

O
CE

O O
CL

X <

o o
I
CL

CD

CONFIDENTIAL
to a servodrive

system

that rotates the controlled

member

so that

It

tracks the local vertical. Similarly, the track control pendulum, sensing
lateral accelerations; (i.e. accelerations perpendicular to the great circle

track plane), produces an output which is modified through the implementation of the lateral

mechanization equation, and the resultant

is

used to

position the missile control surfaces so that the missile remains close to
the

programmed great

circle track.

The

altitude of the missile above the

surface of the earth is maintained by a pressure type altimeter that supplies the essential input to the altitude control system.

Coriolis acceleration computer is used to compensate for the ef-

fects of the Coriolis acceleration upon the action of the system.

Geodesic

acceleration is in general small enough to be ignored.

It

will be the pur-

pose of this thesis to discover whether


tain the

it

is theoretically possible to

main-

errors between the indicated and actual local verticals small


of the local vertical for

enough to permit sufficiently accurate indication


the long-range

bombing and missile guidance problems. Various mechan-

ization equations will be examined, to see which offer the greatest chances
of success.

The most promising

of these will be furthur

examined

to

de-

duce the optimum control parameters.

16

CONFIDENTIAL
CHAPTER H
KINEMATICS

Purpose
It

is the

purpose of this chapter

to

derive the fundamental kinematic


I.

equations describing the motions of the missile postulated in Chapter

The

principle geometrical relationships are illustrated in Fig H-l, and are shown

symbolically in Fig n-2.

The following simplifying assumptions are made


matic equations:
a.

in

deriving the kine-

Lateral and longitudinal motions decoupled

b.
c. d.

Constant altitude flight path


Spherical and homogeneous earth

Controlled of missile

member
of attack

located at center of gravity

e.

Zero angle

and side slip angle

The decoupling

of the lateral

and longitudinal motions divides the prob-

lem

into one of control of the missile in the direction perpendicular to the

programmed great
the

circle and one of indication of the distance travelled in


circle.

programmed great

The kinematic equations are therefore de-

rived separately for lateral and longitudinal motions.

Since the flight path is to be considered as having constant altitude above


the surface of the earth, the only vertical

component

of acceleration is the in-

ertia reaction acceleration of gravity. With the earth

assumed

to be spherical

and homogeneous, this component is a constant over the surface of the earth,
with the numerical value of 115,920 feet per minute per minute (equivalent to
32.2 ft/ sec
).

The assumption of a spherical earth also eliminates any geo1

desic acceleration terms from the kinematic equations (ref par


17
.

App

A).

nJAL
CO

UJ

Z < <0 D >


cy<o
-I
l&J

kJ

z u < o > 5 I o a -I

o< z
h-

o 2 < z z
i -

(* Hi

o < < -I
Hi Of

Of

J < a

a Q I- UJ ui H o < U O o <n

s<
t>

(0

'

CONFIDENTIAL

Vr4)(dvn)

TRUE VERTICAL
l(md)

DYNAMIC VERTICAL

INDICATED VERTICAL

APPARENT VERTICAL
C)(VHJ INDICATED VERTICAL CORRECTION
(PARALLEL TO DIRECTION INDICATED BY PENDULU^"

-(DC)^rue)
TRUE DVNAMK, CORRECTION
DYNAMIC VERTICAL CORS
(DC)(,nd)

INDICATED DYNAMIC CORRECTION

Ql (ombres)
8ESULTANT ACCELERATION OF CONTROLLED MEMBER WITH RESPECT TO INERTiAL SPACE..
t(crrtK rescind 1

9ft e)

INERTIA REACTION

ACCELERATION JF GRAVITY

NDICATED RESULTANT

ENDULUM

ACCELERATION OF TROLLED MEMBER


~Ql(cm)(hor)

CON-

HORIZONTAL ACCELERATION
CF

LCNTROLLED MM3Ee

COMTK'JLLED

MEMBER

I (c

my vert}

VERTICAL ACCELERATION Cf CONTROLLED MEMBER

<*Uc~; ACCELERATION OF CONTROLLED MEMbtR WITH REdPECT TO INERTIAL SPACE ; ', IMPOSED OF ACCELERATION OP CONTROLLED MEMBER WITH RESPECT TO EARTH N ^LUDINO GEODESIC, e HC R _:nta\_ CUCVA~JRE, ACCELERATION) AND ACE'-ERAT^N OF
{

COR. IO LI 5

Q (Cor)*6(GEO)

ACCELERATION OF CO<iOLi_> PLUS GEODESIC ACCELERATION


TR'JE vrv-TlCAL -DIRECTION) OF INERTIA REACTION ACCELERATION OF GRAVITY AT CONTROLLi

MEME"^.

DYNAMIC VER T ICAL- DIRECTION

OF VECTOR RESULTANT OF INERTIA REACTION ACCELERATION OF GRAVITY, ACCELERATION OF COR.IOLIS AND GEODESIC ACCELERATION AT CONTR;^_EC 'MEMBER..

APPARENT vERTICAL-OlRECTION OF RESULTANT ACCELERATION OF CONTROLLED TO INERTiAL SPACEC


INDICATED VERTICAL-DIRECTION FIXED TO CONTROLLED MEMfoEk.
-

MEMBER

WITH RESPECT

ELEMENT WHEN
STATIONARY.

PAR ALLEL

ITS

OUTPUT SIGNAL

I?

ZExO WITH CONTROLLED MEMBER

TO DIRECTION OF PENDULOUS

FiG.n-a. GEOMETRICAL RELATIONSHIPS AMONG DIRECTIONS ASSOCIATED WITH THE DIFFERENCES BETWEEN THE INERTIA REACTION ACCELERATION OF GRAVITY AND THE DIRECTION OF ACCELERATIONS ON A MOVING BASE.

P>

CONFIDE*- {ML
i
i

Coriolis acceleration computer is included in the fundamental guidance


I~5),

system (Fig

and

is

assumed

to

work

ideally (ref par 2

App A

).

The

kinematic equations, therefore, do not contain the Coriolis acceleration.

With the controlled member located


sile, the

at the center of gravity of the

misall

gimbal isolation system can be considered to remove entirely

effects of the roll and pitch of the missile.


trolled

No accelerations reach

the conits

member

as a result of angular accelerations of the missile about

body axes.
Considering both the angle of attack and the angle of side slip as zero,
the velocity vector of the missile always lies along the longitudinal body

axis of the missile; that


is

is, Vj/

lies along Xj- (see Fig n-3).

This

taken as the control direction (CD).


first.

The kinematic equations for lateral

motion are derived

2.

Kinematic Equations of Motion Perpendicular

to

Great Circle Track *

The geometrical elements


illustrated in Fig n-4.

of the simplified lateral guidance

problem are

Figure II-5 defines the directions and angles that are

used

in the derivation of the kinematic equations.

As can be seen from Fig

II-5, the following relations

are true:

(C^A ( vC rtxtc) = (DC)ctrueUtc) - (DC)(

(_

ncj Htc)

(n-i)

ton(DC) (traeKtc) =

g^
20

(n-3)

This development

follows that of John Hutztnlaub, Notebook, July, 1947.

CONFIDENTIAL

Ld

O O
UJ
en en

Q CL O

m
i

21

CONFIDENTIAL

2
UJ CD

O
S z o u

<
a:

O
UJ

UJ

U
<
cr

UJ

O UJ O

^2

CONFIDENTIAL

CORRECT
POSITION

Z(cm)

(C)[(Vert)(op p)(tc)3 cindHDQcindKtc)

(opp)rtc)l(trucf^\true)(tc)

FIGURE IE- 5.

TRACK CONTROL DIAGRAM


2^

^lE-(cm)Htc)

VLE _ Cair)1(tc) - V L(aiP) . M] sinA 2 (crn) cos (C)A (Vert)(tc)


.

(n-4)

From

(n-2) and (n-3)

R E (C)A (Vert)(tc) r -9lR tQn(DC) (true)Ctc)


(H-5)

But, from

(II- 1)

tan(DC) UrueKU)

= tan[(aA (VerlHtc) + (DC) (indKtc) ]


tan(C)A(VcrtKto
+

tan (DC)(inch(tc\
to-ri^DOandiao
(n-6)

1- tan

(C)

A( Ve rt)(to

Therefore,

P
<C)A (VertXtC)
IIR

ton(C)A CVert)(tc)
l-tan(C)A (VertU tc)

tan (DC) (Cn(

)(tc)

tan(DC) (indKt etc)

l-tan(C)A CVertHtc) tan(DC) (tn d)(tc)


(H-7)

This

is a theoretically
if

correct expression,

if

no vertical accelerations are


to

present, or
gjr,.

vertical accelerations are


if

assumed
\

cause a variation in

However,

the size of (C)A/

ver +\/* c

does not become greater than 10


in 30,000 is introduced

milliradians, an error of no

more than one part


tan(C)A/
If

when
all

(C)A/

.v

is substituted for

vert w+ c y

This should be true for


(II-7)

cases of practical interest.

this is done,

eq

becomes:

34

c WH>EJVr/AL
C A lVgrtXtc)
>

ir\A

9i R

nd)(tc>

9ir

tQn(PC) anduta
l-(C)A (VertK1fc) tan (DC) (uxi)(tc)
(n-8)

~^R E

'

If

-s~-

is defined

as

W^,,

then eq (n-8) becomes:

IWMW
/rvA

+ W* -1

(C)A (VertU t
l-CC>A (Vert)rt c, ton(DC) (indKtc(

* w" E

tQn(DC) (lndutc)

l-^ A (Vert>(tc)

ton(DC) (tndutc)
(n-9)

If

(DChindHtc) * s

sma ^ er

tnan f^teen degrees,* (DC)/.*!)/*-) can **


for first order theoretical solutions.

substituted for

^(DC)/.^/^
an
(^-')(indHtc)
'

Also,

(C)A/

ver t\/f c \

as *** P roc* uct * two small angles, becomes

very much less than unity, and can be neglected. Then:


(C)A (Vert)(tc)

W nE

(C)A (VertKtc) =

~WnE (DC) (ind)(tc)

m-10)

Using the same small angle assumptions, and the additional assumption
that

A/ cm \ remains less than

fifteen degrees,

eq (n-4) can be written

in

the approximate form:

V[E-<cm)](tc) *

[E-(air>](tc>

""

V[(air)-M]

A i(cm)
(n-ii)

Ref par 3 App

A
25

CONF'ii

IA!

Equations (n-2), (n-10) and (n-11) are the basic kinematic equations that
are used in
all the

derivations of Chapter in.

3.

Longitudinal Motion Kinematic Equations *

The longitudinal problem


elements
II-6.

is

one of range Indication. The geometrical

of the simplified range indication

problem are illustrated

in Fig

Figure n-7 defines the directions and angles that are used in the der-

ivation of the kinematic equations.

As can be seen from Fig n-7,


^

the following relations are true:


(II

DC

(trueKlonq)

C ) A (Vert)(!ong) + lDC) C ind)(long)

"

12)

a'LE-(cnrO](lon<j)
A, 'r(true) ~
(n-13)

tan(DC)(trueHlo

1_

TIR

(n-14)

Taking the tangent of eq (n-12)


+~ Tan <r\r\ I UU \ t rue)i lonq)

l-tan(C)A 7~T "XTT

tan(C)A Cvert)(lonq)"' ton(DC) CLadUlona)


3

(Vert) (| onc p

tanCDC) anci K |o

r=r^\

S)

(n-15)

Equation

(11-15) is a theoretically

correct expression but, as in the

lateral case, all practical conditions find (C)A/

ver *\n onff \


to:

smaller than ten

milliradians.

Then, eq (n-15) simplifies essentially

This development

follows that of John Hutzenlaub, Notebook, July, 1947.

26

CONf/DGfii

5
LU

CD

a: Ql

Z o <

UJ

<
or

o
0)

Z 2
LU

UJ -J UJ

< o O o
i

-J

cr -

UJ

OL
UJ

o
CD

e
<D

27

I(cmKres)Uong)

(G|(Vert)(ind>(lonql|

[E-(cml]

Oong)

= (C) A(ver
GREAT
CIRCLE TRACK

t)(| 0nq

f(C)A r

L"QE(crr7)fre5)(lonq)J

(C)[CVert) (Qpp)(lori<3)J

(lnd)

^[(Vert)(vnd)(\on S )] (dep)

Wert) (dep)

2
CVert)^ d)(loncj)

(cry,)

^ er^'f\rue)(\onq)
[(Vert)(irueXlonc^] Cdep)

fis.

n-i

LONGITUDIMAL INDICATION DIAGRAM

28

C N i"
rwrv^ WUfc '(trtttHlwq)
IWfTA

f)

(C)A (VertH onq


| i

^ tan(DC)andH onq>
|

tan (DC^

T~
(H-16)

From eqs
+^%lT\n\

(n-13) and (n-14), - ^r(true)^E.

A r (true.)

tir

w nE

(n-17)

Substituting

eq (n-17)

Into

eq (H-16)
|

Af(true)

W*n E

CC)A (Vertu onq) + ton(PC) (in< )(lonq) ^*^C)A (Vcrt)( oa<1) tan (DC) (Ln j u(on<5)
j
|

(U-18)

From
V

this there results,

as the general longitudinal kinematic equation,

*r(true>

_ yy*

(C)A (VertK onq)


|

nE

f-OA^tHlon,) ta " ^C) andu

ons)

^yy*
'nE.

ton (PC) and)(lon9> l-(C)A (Vcrtulon^ tan(DC) and)( on q


|

(n-19)

When

the

same small angle assumptions are made

for the longitudinal

kinematic relationship that were made in writing eq (n-10) in the lateral case,

eq (n-19) reduces

to

\(true)- WnE (C)A(VertU

on<

^+ W n (DC)

tnd>(lon<j>

(n-20)

19

CONFIDENTIAL
This equation and the relationships

(n-21)

and

A r( app)

A r(Lnd)

+ CDC)

Cin dKloncp

(11-22)

form

in all the derivations of the basic kinematic equations that are used

Chapter IV.

30

CONFIDENTIAL
CHAPTER III
THE TRACK CONTROL PROBLEM
1.

Introduction
The simplified kinematic equations developed in Chapter II

for the track control problem are restated here for convenience.

" I^e A Cfc-(crnfl A Vert) ftc) *W^( _<> w,m (tc) (+ r *e lw


r .
\

(m-i)

(QA(S/ rtXtc)

+Wut
'

(CJA^eHrXtc)

-""*

Wa. (DC)6 d)(tc)

(IH-2)

*CE-(cm)](tc)

"

Vts-fdirflftc)

"Vcrair)-M\3

A item)

(in-3)

Using these simplified kinematic equations, various mech-

anisation equations are investigated in an effort to determine a


physically realisable and reasonably accurate mechanisation of
the track control problem. As previously stated, for the purposes

of this study, the track control problem is considered as de-

coupled from the longitudinal problemfl)^ and vertical accelerations are neglected! 2 ). It is further assumed that the missile

operates with perfect control surface servomechaniSms and that

1 - ref par 2 App A


2 - ref par 4 App A

31

CONFIDENTIAL
it

possesses perfect aerodynamic response^

'.

Finally, the pendulous

accelerometers are considered to indicate instantaneously the direction


v of the resultant acceleration

'

The purpose

of the track control

mechanization

is to

reduce toward

zero any angle (C) [(V ert )(

motion of the

mcn}(tc) ^ a* nas l)een introduced by lateral missile away from the prescribed great circle track, and
small to insure arrival within one
is

to maintain this angle sufficiently

mile of the destination.


signals received

The control

achieved through operations upon


of the available angle
in

from

the

measurement

(C) [(Vert)app]/..x, the

pendulum angle,

order to correct the head-

tag angle,

A
[

X(cm)

(CD)]"
it

Throughout the derivations which follow,


to Fig II-5,

is

convenient to refer

which indicates the relationships among the important angles.

2.

Proportional Control

The simplest possible mechanization


which the heading
of the

is

proportional control, in
in linear relation to the

missile

is

changed

pendulum angle as measured between the pendulum and the controlled

member. This case

is

therefore considered first.

The mechanization

equation for this concept can be expressed as:

Azccnv*

= ~

S LC(C0)][AA] (DC) (Ln dxtc)

(m-4)

(3)

ref par 5

App A

(4)

ref par 6

App A

32

confidential
Using this mechanization equation, and the simplified kinematic
equations, numbered one through three, a performance equation
for this system is now derived.

From eq (III-3),

A-

ME-foir;3ftc)

ifcmF

VcE-fcm)](tc)
(

"5

MTf'iilO-M)

Integrating eq (III-l), assuming the constant of integration as

zero^,

and substituting into eq (III-5),

MJ

From eq (III-2),

HbCWr

fC

^r

Ktc)

-HC)A (vJ Utc)

(m-7)

1 - par 7 App A

S3

LUNHD;:

,
i

Substituting (Ill-b) and (III-7) into the mechanization eq (III-4),

Vz.E-(&>r)mc)

*- V~'r*(sjert)(tc)
* [fair)-M3 Tfair)-

_ = 5

CcfcDWCAA)

(Vert)(tc)

(iii-b)

From this, the performance equation can be written:

E.W. 'NE E

Kairi-M]

5 ^[c(cd)]Caa3

^)A

rvert)CtC

)+

Ne

(C)A (vert)(tC)

LE-(&ir)]ftc)

~
TcCcdHCaA}

V foir}-M] VT(Zir)

9)

Although this equation provides a stable control system, a


forced error results from the lateral component of the velocity
of the wind, V
_
(

a i r )]

This, obviously, is unsatisfactory

for the control of guided missiles, because no large steady-state

errors can be tolerated.


3.

Integral Control
Next, let us examine a system that changes the heading of

the missile in accordance with the integral of the pendulum angle. The equation for this mechanization can be written

"

ifcm)

ccfcD):i[>

^ /(DC)

dK+c)

dt

(111-10)

^Ohr

lOi

NiiAiL

Differentiating (III-6), and substituting with (III-7) into


(111-10) differentiated,

VC6-fair)]ftc)

^E (wAft/erQfo:)

_c

(CJA(ygrt )ftc

^.v

(111-11)

or

(r\k

^' M (Vert)(tc) -I
'

-^
a

fr)A W/
^

VVert)(tc)

= QIC

\/ V

[E-^r)]('tc)

+
.A]

(|||-12)

Vc^

ir) .,,

This system, although the forced error is now due to the

acceleration of the wind, has no damping term. S[(c)(CD)l

[AAJ

can be made large to reduce the size of the forced error, but

this results in an undamped period of oscillation of approxi-

mately 84 minutes for the controlled member. Any residual forced error (since it is physically impossible to make S("/qwq D
infinite), causes this oscillation to appear.
4.
)1
r

A jn

Second Integral Control


In order to reduce the forced error, consider next a system

that changes the missile heading in obedience to the second integral of the penduluia angle.

Here, the mechanization equation is:

'Z(cm'

^-^co^

ff(DC\^ Htc)

dt dt

011-13)

d.Z

CONI
Differentiating (III-6) twice, and substituting with (III-7)
into (111-13), differentiated twice,

Vfc -foiWtri

(C) f\ (s ert )kc)


,

(C/A^ ert )(tc)


|

(Yfcir)-M]

\.

( r \*

CI

,_14 )

UE

or,

ff^A

-US

VcCd.r)-M3

/>W

<

V[L< A ir)-M3

/fU

L
^e

\/

ie

'^6

By Routh's stability criteria^), this is unconditionally-

unstable.

5.

Derivative Control
In order to give a complete presentation of possible types

of mechanization, consider the heading angle to be controlled in

accordance with the derivative of the pendulum angle:

A*,c*o dt

= -b LC(cvnck ^

(DC) 0Vw0ftc)

(iu-16)

1 - Routh,

"Advanced Rigid Dynamics".

38

CO,
Differentiating (III-7), and substituting, with (III-6) into
(III-lo), differentiated,

coa

(Vert^tc)

^(wA^

/*

)(tc)

(iu-17)

From (111-17)

(C)Av + W* w NE ^^M vV .^c)T


^

ll
I

H -r

fefco)]CAAa

V rfair) . Ml J

l/hA (UA^ )uc) ~ Yli_^lM

^
v>rc.'c )]lAA3

V [^)-Mj

By Routh's stability criteria for a cubic, this is uncondi-

tionally unstable. Moreover, there is

forced error due to the

velocity of the wind.


6.

Pr oportional Plus Integral Control


As shown in paragraph 2,

proportional control produces a

solution which, although it has a forced error caused by wind


velocity, is nevertheless stable, while integral control (para-

graph 3) develops a solution that does not have any forced error caused by wind, but which also has no damping, and is therefore on
the verge of instability. Thus it might be expected that by com-

bining both equations, a stable system without forced error due to


velocity of wind might result. The resulting mechanization equation is:

*<cm)~

-Wd)]Caa]

(^M-.'ndXtc)

^CcCcd)]Iaa3

^W(^d)^c) Clt

(111-19)

CONFIDENTIAL
Differentiating eqs (III-6) and (III-7) and substituting with
eqs (III-7) into (111-19) differentiated,

=5
From eq (111-20),

CO A (\iert)(tc)
[Cfco)3CAA3

w: Hi

+(c)k fvert)(tc) + 5 cc(cd)j

;(oa
taa:
J

\^j

^+(C)A
ME

(Vrt)^c) fVt

(111-20)

(C)A(^^hc) -\-

vv NE
CcfcDl](AAj
il

2
V^a.ir)-M]

4-

OccftoD Caa]

w;

(C)Afv ert )(*c)

+W

WE (C)A(vert)^tc)

H~W we
>5rcfcD)][AA]
V[&i7)-m] (CD)] [A A] VrraiV) -r

VO

fiirfl

(111-21)

By Routh's criteria of stability, this system is stable if:


* M<i

to StfofcoflCA*]

5 lwfcoa rAA]

pMitive numbers,

(b)

>0.
v Ralr VM3

~T<c) (c oi] Ca a]

Furthermore, as expected, this system has no forced error

resulting from wind velocity, although it has one caused by the

acceleration of the wind. This suggests adding second integral


to the system just discussed.

88

CONFIDENTIAL
7.

Proportional Plus First and Second Integral Control

A ^=-5fcfeDW x3CDC);
tfc

ni)(te)

ttfcBOt

Afc/ftW D(wd*
tf

S fcrcoi]DkXj //

(DC) rmd)w dt dt

(III -22)

Substituting eqs (III-6) and (III-7) differentiated twice,


eq (III-7) differentiated, and eq (III-7) into eq (111-22)

differentiated twice,

VL.m-K.O)A

fort) fee)

~ ^[cfeomX M

5TT VV
WE

"

lWM fo*)fcc)

-f-S

/W,
i

fcl

fc)A fortKtc)

w:,

roA

(111-23)
(VertX-tc)

NE

C('i>")-M]

From eq (111-23)

9*
S V ^Cc(co)3 {a A3 v O-)-nO

4-

D^CA A]

s '-'ccfcoatA A3
{

(OK^r

w;+

'CcIcoMCaaj

(CXk
J

^cfe P)JCAX3
I

vjl

r^A

3ccfep)JCAA3
'Ccteo>3

v^* /)a

=-*'lcCcD>3

V
CAM
V CC&lr)-MT)

^,3

(HI -24)

S ^tfcOUCA A]

scpfcpQCV A3

Ca aJ

According to Routh's stability criteria for a quartic, this


system is stable if:
(a)

The three sensitivities have the same sign. (They may be all positive or all negative),

(b)

i.

NE

Ccfc^CAA3

^Lc-fco)

q s

Laaj

....

V
\/

'

^Cc(co)3t>A3

^^

tfa*ii-)-M3 Mj

39

CONFIDENTIAL
In this system no effect of the wind of lower order than the

second derivative of the wind velocity enters into the forced


error. It seems unlikely that a forced error of this nature could

have a magnitude large enough to be of practical importance. However,


if this error should prove important,
it is possible to go

to higher order performance equations. One further case is con-

sidered.

8.

Proportional Plus First, Second and Third Integral Control

Ancm)

^ccfcouo; vn'DQ.'rvnfte)

S [cfcD)]C x

a]

/(

DC)r nd)(tc) dt
.

5 Cc(co)K ;^

/TDQ, dXte)

dtdt 5

[cfc paC A>;j

//

MDQ

;h< j )fe)

dt

dtdt

(MI-25)

Substituting eqs (III-6)and (III-7) differentiated three


times, eq (III-7) differentiated twice,
eq

(III-7) differentiated,

and eq (III-7) into eq (111-25) differentiated three times,

(OAtwVtc)
v^'^s/trdbdl
^fcfcoa ca a]

(*ir)-M]

w:

fcplJCA a'3

\@L^+ w
1

(C)

V * WE

VVe^Xtc)!
:

)(C )A rve ^
-TcfcDfl TaaJ

Vtc)

w:

<C)A /Vert)

(fa)/

(111-26)

From eq (111-26)

(C)

A^^fcsf

^-jos^ CC)A ^-f |w^ 5cCve


.

>j :

a'j

fc)Arve^ fe)4-

Wi

OccCcpOCa

'A"J

ass)
5*

S
fc)AvtrtW +fw;

CctcOlirA Aj

(C)
^tcfeolltt'

AV*

tj

+ {VnT

Ccfcp)3CAXj

)-M3

JtcCcpOC' A A}

"^fr'cDOCAADy

\(i 'coi:

ca a)

sy

"CcrooacA

}/C)A AV

NE

.-.

(IM-27:

S
C^coDC'X Aj

\l

VCTv'r)-Mj

V I>-<W3

40

CONFIDENTIAL
By Routh's stability criteria for a quintic, this performance equation is stable if:
(a)

All coefficients of the homogeneous equation are positive,

(b)

5 [C(CD)DCAA1
TCtCPflrAA]

w:f<3

:c(co)3L'A'

a'J

'CcCcdBC a

AD

+
wi l
'NE
j

S
s

(c(cp)3 L 'a

a3

IB.

+
*r(i!r)

^Cc(c O^tAA)
'Ccko)] CA A3

^cc(cp)3Ca a 3 \
'CC(CD)][A

/\a/

^[c(cp)KaaJ

^rcfco)DCAA]
'CCfcDDCA A]>
..

N6
A*J>

^CctoD)] [AA3

M)

5 [c(cd)DC'a

[cfcpUCA A],

IR.

-.. 5u JMc(cdV)Ca a3

V
V

CcfcoHtA A]

s
'^(co)3C'a"a'J

'|R.

S V ^CctCPUlA A3 V/ 1
)
I

s ~Ccdo)3CAA3

5
E

5 Cc(cd)3Caa3

IR

'frfcp^

[(,r)-M3

S ^Cc(cD)D[aa

5 C^coUCa

A'y

"^

Cc(c 013

La

'X]

Ne

v K4 ir)- M]

5,[c(co)3L'a"a]'

s
Cc(co)3C'a

A3'

Cc(CD)3tA'A'3

[Ql '

M3

Tcfcp",

These restrictions permit the sensitivities to be either


positive or negative, provided that they all have the same sign.
It is of interest to notice that in each stable performance

equation, the forced error has the same type of coefficient, with
the order of the wind derivative increasing from velocity for

proportional control to any derivative desired, as the performance equation increases in complexity. This coefficient is of
the form:

^CcfcPQEA a J
[c(cdCa a

*^ rcfco)JCA

a'3

U A
S
S

^C= (cd)2 C a'a


[cfepUCA a ]/

'Ccfco)3[A a]

V w v /
V

s
E

rv
s
aJ
IR

'frfcoQCA A3

>0
IR

5 Cc(cW3['a

S ^fcteDXitTk'A'J

'/

S UCfCD^[*A"A3

Cc(cp3]t'K

1/ V A]riAir)-M3 CA

+
'

s,Cc(cp)3Ca A3

'CCfCPnCA A]
CCfCD)3tA A3

Q
A], [cfcp)3UAi

It

CONF/OtNTJAL

me

<cd)0CaA)

The forced error is thus seen to decrease as the missile airspeed increases, and as the sensitivity S[c(CD)] [M] increases. Theoretically, if this sensitivity increased to infinity, there would be no error, either transient or in the steady state, caused by the action of the wind. This is similar to the conditions found with integral control, eq v-lxa . i.c whor* a* +h* ^m (111-12) wnere, as the . sensitivity
> i

increases to infinity, both steady state and transient wind errors reduce to zero.
[C(CD)] [AA]

ra ck Control Corre ction as a Function of Wi nH

Since the mechanization eq (111-22) for proportional plus first and second integral control appears to provide a stable system without excessive forced error, this system is considered further. The performance equation for this system, eq (111-24), can be expressed in operational form:

* NE

'WA(v rt )ft e __
) .

V -tc(CP)JLA*]

<S

w
CCa.ir)-Ml

r
J

(111-28)

vy cccc D)3 ca xj Vr &ir) . M1

o wCDI7r vvJ'
r

.....

<\

ne NE
....

(r r

C c
,JLAAJ

W.. W6

'

S>

CO.
If (C)A/ vert

<

is expressed in minutes it is a measure,

in

nautical miles, of the transverse linear distance of the missile


from the desired great circle track. The oresence of the operational symbol in the numerator of the equation indicates
that

there is no error, in the steady state, from an accelerating wind. If it is desired to find the missile response, A2( cm ), to
an accelerating wind, it can be calculated if eq (111-28) is
A Z(cm)
-j~*J,
(
'

multiplied by an expression for

The result

'

A (vert)(tc)

yields the heading of the missile as a function of wind velocity.

Manipulating eqs (III-l), (III-2), (III-3), and (111-22), and


expressing the result in operational form, fs

OC'aaJ
<3
IR

vn 4

^[c(cpT]^AA3'^-t

(crrO

P
IR

'CCCCD)J

CAA] >^E
'

^[CfcPVjtlA A)|

W[

ct )][M]

ptOi

'CcfctrtHAAj

IR

(111-29)

(C)A (vert) (tc)


Combining eqs (111-28) and (111-29) gives,

P
i(cm)

+'

fs Cc(cpDCaa3

<^r6r)-Ml

s
IR

...v

fP^
"

'[CfC0)KAA3
'lc(co)ll A A]

W
V

V[ (air ;-m3

:J>
;

CcfconcA a]

WE

P
5

W
\J

Lte'ir)-KL

^CcCcd)]C'aa3 ^C(a"^-M]j

S
NE

(HI

p*+ p

*^ rcfco)3CAAl

*Ca'

P +|w-+ T

-]P +

fs

CD)]CAAJ

8*

CONFIDENTIAL
The negative sign shows that the heading Az(cm) of the air-

plane is correct to oppose the tendency of

Vk

. (air)]

*o blow

the missile off the great circle course. This is true even if
the sensitivities are negative.
10.

Track Control Closed Loop System


The relationship of eq (111-30) can be obtained from a study

of the over-all lateral system as a servomechanism. In the direct

servo closed-loop system, the angle (C)A/ vert

does not appear,


)

but it can easily be obtained, as indicated in the accompanying

Pig III-l.
In this figure:
(FF)i has the value

SIR

(PF) 2 *

zi
'

CM )

DC (ind) (tc)

and is made up of the sum of two performance functions,

(pf). -

(
.

llnd)(tc) ^! DC >(true)(tc)

(PF) D m

Z[cm) ( DC ) (ind)(tc)

which is the mechanisation equation in operational form

(PF) 3

air ) = M J ) p> SIR providing the proper function for the comin the feedback path, is (t(

binor, which is physically the pendulous unit.


44

CONFIDENTIAL

2
LU

>

tn

<
M

<

>
a o o
o
UJ in

O
-J

>
in
o:

o o o o <
Ct
I-

ix.

45

H_-

fV it

AL

The pendulous unit, if (DC)( true ) remains smaller than about fif-

teen degrees, solves the equation

<3, R

'

"^^"^
(III-2),

(,,,-3,)

This equation is derived from eqs (III-l),

(III-3)

differentiated, and the relationship that

ocM = rc)A
Fig II-l.

(vfert)fte)

+ (oo

(ini)M

cm -32)

This last equation can readily be obtained from an inspection of

Solving for the performance function (PF) a


(III-2),
*

using eqs (III-l),

(III-3), and (111-32),

The mechanization equation, which in this example is eq (111-22)

expressed in operational form, is

fPFl=

M^

4> -rfesaga

p4.

^fc fcp)3I >*H

(111-34;

cm)

P
46

tt^Ud)(ic)

^UMi

..

The product of eqs (111-33) and (111-34) gives

s
(PF)

...

r<3

A
(DC)ftruueKtc)
1

(in

-35)

but, from the theory of servoroechanisms, (1)

A., cw)

(PF), fiPFj
i

si*-**

+ rpFi cpf)

(in-36)

Solving this equation gives eq (111-30), as before. The stability


criteria for eq (111-30) are identical with those for eq (111-24).
Since the sensitivities can have either sign,
a

necessary condiSIR

tion for stability is that

S[C(CD)][AA]

>

V [E - (air)]
If this is expressed

if the sensitivities are chosen as negative.

as

...

=-

'|R

(wWeK>ia*<15<0,),

(Mi-37)

air)]

1 -

Brown and Campbell,

"Principles of Servomechanisras".

COWFIOfNTlAL
then, for stability,

K>

c
CcfcrtUAX3

(1

11-38)

11. Numerical Solutions,

Using Two Quadratics

Using the relationships that have been established, and

substituting numerical values for the coefficients, it is possible to solve for the various angles as a function of the accelera-

tion of the wind Vrg

(air)]

From a sequence of such solutions,

using a range of values for the sensitivities, it is possible to


select the most satisfactory values, and to estimate the effec-

tiveness of the system. This is now done.


The homogeneous portion of the performance equation, (111-24)
is a biquadratic. It can then be written
x
1
)

(p

+ 2(DR)W, +W,

(p*+z(DRl

t )

(111-39)

which factors the equation into two quadratics. Manipulation of


the equation indicates that if the damping ratios and natural

frequencies are varied, the higher damping term associates itself

with the shorter of the two frequencies. This result then appears
essentially as that of a lightly damped long period quadratic,

which is undesirable. For this reason (DR)^ and (DR) 2

and W^

and W2 have been chosen equal for all solutions of the equation, 48

CONF/DEM IAL
and equations art now derived to facilitate the selection of

values for the sensitivities to make these limitations hold.

Expanding (111-39), in the general form, gives

p +[2 (DR)W+Z (DR\W] p + [W,


a

+W

MO^D^y^p^
(111-40)

[2(D^yv +z(DR)xw;]p+w;w;. =o
if

w2

%E

*nF

(FR), and (DR)! - (DR) 2 - (DR) then,

comparing the coefficients of eq (III-40)with those of eq (111-24)

gives the following relationships:

IR

+
V
Cftir>-Ml

Ccfcp)]CAA3

- 4 0>R)frR)Wuc

,
(
I

Cc(cc]

CX XI

C^co)3[XX3J

-41

s
ME

zrFRTw; [2(DRr+i]
s
)

(111-42)

^CcCcorjCAA

|^W;.= 4W(DR)WI
Ccfco)3CA A]

(111-43)

5 W>
CcfcoTKAA}

(FR)

(III -44)

49

'

CONFIDENTIAL
When these four equations are solved simultaneously for the
sensitivities and for the damping ratio, the results are:

M5wcoBC

(fRf~

2(FR)
<3

(111-45)

"J=

iR

8W

V BE

l 3CDR) (FRr
x

"

_46)

t,

" atiX!

2V 0MPMfl

fl>R)

(,,, -47)

a**,,*

8V^,. M]

(W

(l,| -48)

It is noticed that the choice of a damping ratio immediately

fixes all sensitivities and the frequency ratio (or that the choice
of a frequency ratio, alternatively, fixes the damping ratio). As
the frequency ratio increases; i.e., as the quadratics have pro-

gressively higher natural frequencies, the damping ratio simultaneously increases without limit. Conversely, as the frequency
of the quadratics approaches that of the &4 minute period "earth

pendulum", WNE

the damping reduces to zero. The sensitivities,

at the same time, increase without limit.

CONFIDENTIAL
12. Numerical Solution

Using Four First Order Terms

A second method of writing the quartic equation, which can

be examined easily, and which may yield results of practical


interest, is that in which the equation is reduced to four decaying exponentials. This equation is of the form

(p

^,)(p

o)(p a)(p

+
(F).

(111-49)

When expanded, this becomes 3+ + [(cT) (cr)+<cT\(cT\* rcrycT)/ (ct\(ct)* (ct)Jct\ P^kcT) (cT) fcT)Jp 3

Vr)/
a~

"

"CCT)/CT)Jp
.
!

[fCT\(Cr)t (CT\

(CT\(CT)X (CT)* rCT)/CT)

(CT^fCT^ (CT^CT) J P
(111-50)

(ct\(ct\(ct)3 (ct)+

The general relationships among the coefficients of eq (111-50)

and eq (111-24) give

9
ycc fco>] CA
*S

A3

V v

+
MJ

S
s "* Cc
(

Cfai r)_

co

V}

za

a3 _

(tD+CCtt+CcT), -KC'TJL

-51

(MI-52)

51

CONFIDENTIAL

(Hl-53)

CCfCOWCA Al

"

(CT}(CT\(CT)JCTl

(111-54)

The special case which is considered here is that for which

(CT)i (CT>2 - (CT)^ - (CT)j^ - (CT). Under these special circum-

stances, the simultaneous solution of eqs (111-51, 52,

53,

and 54)

yields

V^(GV5z), (bf z
from which
1

(in-55)

cct)

2.418 =W. 'MC

or^=W
fcT)

Q4IZ.

(111-56)

CctcoXlCA X3

ie

w;.

LfeW-M]

4w;
CCT)
I

4
:

(m-57)

(CT)

X
I*

(CT)

(mi-58)

CcCcoHCAA}

VO^lD-M]

w;52

(CTl

CONFIDENTIAL
I

(Of

C(a-)-Ml

4W*

(IM-59)

(at

Evaluating (111-57), (III-5S), and (111-59), using eq (111-56)

with the smaller (CT),

IR.

TcCcoUIaAi

M^vM3WMt 45
<3, R

(111-60)

1.205
(
1 1

TcrcD))CAAD

V
<3

-61

W
(III -62)

CcCco)DCaa3

M,;^0728

This gives negative values to the sensitivities. The alternative evaluation would make the sensitivities positive, but would result in characteristic times fdr longer than could be tolerated. The system, with negative sensitivities, is stable by Routh f s
criteria.
13 Numerical Solution Using Quadratic and Two First Order Terms

The final method of examining a quart ic equation which may be

^3

7
CONfiDhN
I

IAL

of interest is that in which the quartic is broken up into one

quadratic term, and two decaying exponentials,

(p'+Z(DR)Wp +

W )(p+ ^: )(p+ g-j)


t

(in-63)

Upon expanding eq (III-63) this becomes,

rgfoiflw lfcT),fcT^
.

W.ji/'Ib
fcT)/ fcir^Jr

(111-64)

The general relationships among the coefficients of equations


(111-64) and (111-24) are

k H? l^Kfc.ntAi] V Wr) M]
.

|sfeaAii]
5ccfconr>iA]J

= 2 cDR )w+-L.+J)

lCT7 t

(111-65)

5
5

--^wn

+
(cT),

ca)/
(m " 68)

W" = <tf)(CT)
54

com

These equations can be solved simultaneously for the sensitivities, and for the relation which connects (CTJ^
,

(CT)2
(FR)

W and (DR). It is convenient to use the relationships:


W NE

(CD^E
[(CT)PR]
2-ir

and

[(CT)PRj 2

(CT) 2 WNE
.

The

2-rr

resulting equations are:

41T (FR) |YCT)PR] [(CT)PR] +

= 4TT

[(CV)

PR], [(CT)PR\ + 4TT (DR)(FR) ([(CT)PR], +[(CT) PR],

I ;
1 1

-6?)

ZTT
fc(CO>]

Q x fFKV

[fcT)PR],

[fcT)PR] t
C 1
1 1

[A A]

H,lir)

M]

(4Tt (DR)(FR)[(CT)PR], [(CT)

PR\+ir(m

([fCT)PR], +[CCT)PR] 2 )-(DR)(fr)-TT ([(CT)PRL], +[fct)PR],')}

-7

<3
Cc(co)]

f(DR)(FR)-Hr([;(CT)PR]

+ [fCDPE].

)}
IT

CAa3

V.

M]

[4TT (DR)(FR)[(CT)PRj,

^CDPRl + TffRy^CTjPiq+pCDPRlJ-COR^FIZ)<3

([(CT)PR],

|fcT)PR|,j

(M

1-71

w.

rcfcDOtAX:

VL(o-nO

(8trYDRXFR)LCCT)PK], [fcflPRj,.

+ 21T

1 fFE) ([(CT)PR],+RcT)PE]>2CDR)(FR)- 27T ([fcT)PE],

+ [fCDPE],)];,,,^)

CONFIDENT^
These are evaluated for the single case of (DR) If
[(CTJPRJ!
,

with

[(CT)PR] 2 - 0.1. Then, from (111-69),

(FR) - 0.307

This gives:

SX*,

y^

Wms

3764

(||1 . 73)

9.J.205

^crcD^cAAj

9,. vi

Hi

0.714
(III -75)

^OairV-MiJ

Notice that once again the sensitivities are negative.


14.

Plots of Results
It is now possible, using various coefficients for the

equations of paragraph 11, or the indicated coefficients of the


equations of paragraphs 12 and 13
,

to obtain numerical answers

by solving the equations with various wind inputs* This was done
on the Rockefeller Analyser at Massachusetts Institute of Technology.
It is not necessary to solve for various missile speeds, since

the speed of the missile can be combined with the heading angle of

the missile A z ca ^ j

This is possible because of the linearisation

assumption which was made in deriving the kinematic equations, and


56

CONFIDENTIAL
holds only if the missile airspeed is very nuch greater than the

velocity of the wind.


Plots were made of the angular error in minutes of arc of
the position of the missile perpendicular to the great circle
track. Considering the earth to be spherical, this is directly
a measure of distance from the track in nautical miles. For

convenience, on all plots, the unit of angle is the minute of


arc, and of time, the minute of time. The foot has been taken

as the unit of distance*

The first input of wind which was used in solving the equations was a rectangular pulse of wind acceleration, or a constant

acceleration of the velocity of the wind from aero to some finite


value. The duration of the pulse was taken as one minute. Next,

responses were taken for a step function input of wind acceleration. Finally, responses were determined for sinusoidal inputs of

wind velocity using several frequencies. Plots of the results


follow, using various sensitivities in eq (111-22):

57

confident;

r^klL

u CctCD)HAA]
i

^[C(CO)]lAAl
[C(CD)][AA]

(DR)
.417

^LC(.CD)HAA] V[( a

r ,-M]

LC(CD)UAA1
0.4195
1.789

0.00000208
0.00001158

0.0281

.750

0.0885
0.2172

1.050

0.00003540
0.0001255
0.0002891

4.890
18.05

1.500
1.875

0.6575
1.4210

35.75

NUMERICAL CONSTANTS USED IN OBTAINING RESPONSE OF TRACK CONTROL SYSTEM TO PULSE AND STEP OF WIND ACCELERATION AND CIRCULATING WIND

TABLE m-1

58

CONFl

*-

RJTi PJ !

UJ

J
v
>

O
o
I-

/
-*
a 2

<n

^ O
a: UJ
-J

^^
/I
Ol-J
'

3
or

UJ
I-

D
5

Z z
UJ

-1

O
1

d
11 11

-i

S 5

1 0>
zinuj/tt
<

Op
Ck
i
\

Nouva3T3oa V
.0

u.

f=

(U

11

O
11

^n
11

o
01

O to z <
UJ <*
CM
jj

UJ

CO

-a-

40

B
UJ

o
CM
1
1

3 O
U.

J3A)y(Q) *^ )(*-A)w/ JiiAl Nl r sii P* (MDVaidiO S311W IVOIlOVN) 319NVJO S3inNIW W(*

59

wim

IUL.H
>

iaL

HI

o
1

1.
1-

-n 3 a,-i2 2
+
n;

Z ? 5
fejVO

a \A
1-

3<0

o
UJ
<n

3*

yp O
U 2 < CO
UJ

z o
ui

5, *

>CO

-SI"QC

O
Z
Z
a:

'K

< 5 Hj
l

*s8j 2 S O Q O w
u.
"

!
FORI

o o
5

a)

<
o
Z

>NTRO

i
^ s
z

"

u.

c
<f>

C)

uu//^

o w

mSUL u. O
UJ

to
0)

o < a>
to
111

a z
D

a:

to

en
1

s
UJ

o
(XDVtfl JJO S31IW IVOIinVN)

O
e> Nl

319NV JO S3inNIW

W(* J *AJy(3)

a.

60

CONFIDtm*-

s
ONTROL
LERATION

o
to

u Z <
or

Ul
t

o
O
10 Ul
1-

< Z
i-

^
UJ

^k

o
Z
:>

O D * Z
c

UJ
</>

s r m 2 u < Q QC c < 2 r
;
-

5 c

i
z o
en

o z < z co o

NOIlVa31350V
in
it

<*

UJ oc v*

-J

U.J
2
H

n_

Z h

to

H O 2 u o ^

IV U.'<M

O
ft

o
n
,

QC

"

QC

z < Q.
DC UJ
1h-

a,

a
o

3 r < 0r

J _

* <

cL

02
UJ
or
<
4

2
Lo h-

c3

us

**0

s 31IN

'

tvour WH)2

JO 5

INI

Nl

M3/

u. <o

61

CONFIDENTUL
s

O Z

U
Q.

K
"

00
N

o
en

O
*

^5
w a
uj

o
UJ
\-

>
cn

CO

>
o 2 O
or

UJ

* z

H 2 Z o
o

o o I z D o O < u.
or I-

O
UJ <n

2 O
0.
(/)

UJ

or

00

(0

CM

CM
i t

<0

00
I

c*

(S3iiw ivDunvN) Dciv

avmoaio io sainNiw
62

ni

(^X^AJy^ <9

>

'

Uttri&JtJ

SU*.
o
00

o
ii

O
a.

2 Q
o
(0

E
CO

<
o

o m
lO
II

2
Ui
</>

8P
Ui

y.

a
<

\H D or Z 2 UI o o i a: o QC H in o z
UJ

Ico >- UJ

2
o

o~ I o
< 5 or

il-

o
O
CVJ

UJ

o
Q.
(/>
u r>'

ui

r00
It

a T
O
CM
<0
CVJ

a:

(0
I

B
CO
I

co
i

CM

<)

O
CVJ

(S3TIW 1V0I10VN) DV 3*103310 dO S3iniMllM Nl


ea

(OIK^AWqx

u.

o z

< -J
8

I-

u
(X
.c

o
to
'

E
CO

"

<
o
to

^
> 5
MUTE

00

II

O o * H 2 uj
2
UJ
-

CxL

>-

o i
z 2
I-

o
*0

z o I u O H

Z UJ Q o 2 c cc 9 o

V <
(
1-

sS 5
*-*

u.

o
UJ
</3

z o
Q
Q.
ex

CNJ

<D

a
o

(931IW IVDIiOVN)

D3V aVHlDdlD JO S31HNIW


64

NlWKV^My^j

LUi
*

:miAL
\3

< -J
D O
at

u
x o.
if)

O
O
ti

CO
"

L-l

o
to
UJ

2 uj p h a H W 0) 5
-

00
ii

D Z 5

2 ^ *
Z
Ui

Dig
5

i/\

a:

II

o o < u. ^ w or
o
U.

O
oo

o CL

a
i

CO UJ

oo

o
(s3iiw

00

00
i

<0

O
CM
i
i

lVDimvN) oav avimaio jo sainNiw


65

ni (*)U ja A)yQ)u-

COHHQB

iaL
o

%
y

<
3
(J

Of

o
CD

< a 5
O
o
ui

g
Ui
ui

a >
L

UJ of i > Z u

<3

It

Z O -

-4

o o

2 z
l-

o
<

at ^
z

II

o
LL

O
UJ

o
CM

</)

Z o
Q.
(A UI a!

00

c*

<r>

<o

<n

in
i

<0
I

CD

N
l

B
u.

(S3iiw

iv^unvN)oav avinoao jo sainNiw


66

ni

(D*)(*

J9

A)yOj

CHAPTER IV
THE RANGE INDICATION PROBLEM
1,

Introduction
The simplified kinematic equations developed in Chapter II

for the longitudinal problem are restated here for convenience:

vDwfinMlonaT

xJT.

~~ (CjAjVertKlo^)

(IV-1)

Arfiiwir

A r (truO

(C)

(IV-2)

Arrapp)

A rM)+(DC;

flnel)

^ ohJ

Using these simplified kinematic equations, various mech-

anization equations are investigated in this chapter in an effort


to determine a physically realisable and reasonably accurate

mechanisation of the range indication problem. The assumptions


are identical with those for the lateral problem* The missile Is

assumed to move with instantaneous response to wind forcing


functions.
The purpose of the longitudinal mechanisation is to reduce

toward sero any angle (C) [(Vert) (ind) (long)] which has been in-

troduced by motion of the missile along the prescribed great


circle track, and by measuring the angle between the reference

vertical and the indicated vertical, to obtain a measure of the


67

<_

location of the missile. The orientation of the controlled member


is

achieved by measuring the angle (DC)

ncj)

(long) between the

position of the pendulum and the controlled member, and by operating on this quantity to position the controlled member with
respect to the reference vertical. The longitudinal reference is

chosen as the vertical of the departure point,


the value of (DC)
(

(Vert (d e p)
)

Since

ncj

(long) is dependent upon the position of


it is more
Ar a ( pp)

the controlled member, as well as of the pendulum,

convenient, in deriving mechanizations, to use the angle

between the reference vertical and the indicated vertical of the


controlled member, than to use (DC (ind (long) directly. A r app
) )

is

physically measurable, since it is the sum of (DC)

j.

n d)

long)

and A r (j_ n(j), which are both known.

It has the advantage of being

independent of the position of the controlled member.

Throughout the derivations that follow, it is convenient to


refer to Figure II-6, which indicates the relationships among
the important angles.
2.

Trial Mechanizations
A.

The first mechanization that is considered is that in


( )

which the second integral of A r app

moves the controlled member

according to A r ind
(

and its first two integrals. That any simpler

case produces an unsatisfactory solution is indicated by the mech-

anization equations for the track control problem. The mechanization equation is

S[c m>]CA]

//

r(iB j)

dt dt

+5

[cfcml5

1]

AR

f ;j)

dt

+o

[c(cw,YJ[Xa]

= A rVivwl
rrtnJ
~

// //

M r(apP

r(app)
,

dtdt ULCJi

(IV-4)
..i

cor,
s

Differentiating twice,

0[ C fcmJ][AA]

r\r(\r\6)

'

^[cCcw)] [aaJ

^r(md")

'

^(cwiQCaaI

^V(md)

- A K&pp)

(IV~5)

Substituting (IV-3) into (IV-4)

Pccfcm)][AA]""

M Arfind)

'

^[c<'cmr)[AA3

AVindP"

^[cfcm)}[AA'J

fi

ni J)

= L/L(

irs<j)|

^3

(lv-6)

which is the mechanization equation in terms of (DC) (i n d) (lone)


and A r ind (
).

Further substituting (IV-1) and (IV-2) into (IV-6),

/
I

l!

W*5

^[crcm)J[AA3
r Ctrue)

^LcrcvyilirAA]

A rVftrue)

Prc(cm) K AAT
| <.

'

c
^[c(cwO][aa]J

r (t

_ Ma )-^/n rw

T
I

^Ce(c rvO] ZA AT
^[cfcm)] [a
a]

(WA

^[cfcm)] [AA]
~t"
0[c(c*vo;j

(UAi

ME ^[CCcmflCAA] ;

l^[c(cmi:[AA)

[AA]

Ov-7)
This control equation indicates that the value for S [c(CM)l
[

AAl

should be unity, and it will be so chosen in further considerations


of the problem.

Even with the position error eliminated by the


>

proper selection of S("(qwcm)1 [aa]


by the rate of change of A r true
(

tnere remains an error caused


r(true) "E
^3 a

).

Since

mea sure

of the distance travelled over the surface of the earth, and because

airspeed is constant,

(IV-8)

I
.

<

This means that the steady state error is a function of

wind velocity, and that this equation is therefore unsatisfactory.


B.

In order to improve this response, the first integral of


*s

^r(app)

added to the second integral in the preceding mech-

anization equation. The resulting equation is

r(ind)

dt<lt+

[cfcmmAAD

yA

r(imJ)

dt +-5 Cc(on

A]

r - nd)

-J

rftpp)

dtdt + 5Cc(cm DCAA3 /


,

A rrap ^dt

(IV-;)

from which, with (IV-3),

(IV-10)

Further substituting (IV-1) and (IV-2) into (IV-10),

J_ S^^.: ^
l

A "rterue)

_f|--L
1

v./

_J c

U
I

"r

true)

=(C )A

+|^^
C

(OA

+^
S
, ..

fc)A r
(.IV-llJ

In this control equation, there is a forced error resulting

from the acceleration of the wind which can be removed if


1 S

is made equal to

7 W^E* Under

this condition, a

[C(CM)][AAj

stable control equation results with a forced error resulting

from no effect of the wind of lower order than the time rate of
cnange of acceleration. The homogeneous equation contains damping,
but has a period of about 34.4 minutes, because the natural fre-

quency of the system is fixed as W 2 NE when the value of S[ C CM }]


(

AA

is chosen.

Bee, use it is desirable to increase this natural fre-

quency, in order to reduce the time during which transient errors are of importance, other mechanization equations are examined in this chapter, in an effort to acquire control over period as well as damping in the control equation.
C.

The third mechanization equation to be considered uses

the third integral of A r(app) t0 move the controlled member) by

operating on

A r(ind)

and its first three integrals. This mech-

anization can be expressed

"rf IN

d)^^^%m)][w]

/ J^rtmd) d"tdt +

O rc(cm)J[ KA3

/ Arfind)<n H"

J [cfcrn)]r^ A] A

r(

nd)

A,, app ,Jtdtdt


(IV-12)

Differentiating equation (IV-12) three times, and substituting equation (IV-3)


into
(

IV-12],

'[Cfc^flCAAJ ^rCmcl)

^>

CC(C^)][AA] "i-fi'nd)

""

^Ccfcm)]CAA] "rd'wd)

^^'(ind)(\o^
(iv-13)

Further substituting equations (IV-1) and (IV-2) into eq


(IV-13),

Cc(cm)]CAA]

Cereal] [A A] 5,
I'lira.e)

Vftrue)

Tcfc^lCAAj

W we

A.
[c(cm)][AA] ^[C(C

(irue)

(C)A r

S* (CW)][AA]
[C(cmVj[AA]

(C)A r

J[ot "*"i
'TcCcwOC AA]

C )A r

(C)A r

[cfcm> 3[AA3

5 [cCcml][AA] fc
(IV-14)

This control equation again contains a forced error propor-

tional to the velocity of the wind, and when the proportionality

factor 3r ,-,][- AA j

i3

made zero, the homogeneous equation becomes


17

unstable.
D.

NFIDENL'/.L
In an eifort to improve this situation, the second in(

tegral of A r app) is added to the third integral in the preceding mechanization. Then,

HI-A^cftdtdt-hS

Rfcm)][AA]

/fa

l<h

dtdt+5

Cc(cm)J cXa^

I hr(\t) c t +- cccc^] U AA]


"

>

rf W j),-

II
-J

A r(app)

dtdt dt 4-5 cc fcmiDCAA]

A r(app)

dt dt
(iv-15)

from which, with eq (IV-3),

cccc^BtAA]

A,^)+5 Ccrcm)3:A

A:

A Killd) =

(DC)

nd))oh3

+S

[cfc>vi)Ji:AA]

(DC)

)long

(IV-16)

Further substituting eqs (IV-1) and (IV-2) into eq (IV-16),

_J [_

^Cc (cm Tic a A]


.

a
'"Vtrc roe)

J cacmncAAi
'[cfcknrjCAA]

NE

5 Cc(cmuLA'A] w

A, ftr j=<OA P+

^'""'
[cCc-cvOKa'A]

(c)A f

+ |^fli(-C
tc(cmUC*AA]

COA r
'Cc(cm,)]L'aa]

(lv-17)

The lowest order of the forced error in this control equation


is

that of the acceleration of the wind. The coefficient of this

*N

term,

however is
L

s fc(CM)I EAA] S CC(CM)][AA]

W2 NE S[c(CM)][ AA]

If

s fC(CM)]

[*A*A]

is chosen as equal to -1

w2
NE

the acceleration .term

disappears and the control equation indicates that the lowest


order of forced error is that resulting from the time rate of
change of the acceleration. The coefficient of this term is of

similar form to that of the acceleration term, and is

72

-OWDENTIA!
1W
S[C(CM)1 [AAl

It thus appears that the acceleration


[AA].
LC
\

NE

[C(CM)]

rate

term can be made to disappear if

CM u

AA

ia cnosen

[C(CM)][ AA]

as

w NE

From Routh's stability criteria for a cubic, the sign of


each

term in the homogeneous equation must be Dositive, and

CcfcmVlCAAl

[cfcmMAAl
[c(cw)]['aa]

>
[CCCW)]LA
A]

(lv-18)

5
[ckmllLAA]

Equation (IV-18)

shows that, with S [c(CM)][


;

*A*A]

" ~~2

the

s y stem

NE

is

stable only if
S

[C(CM)] CAA]

[C(CM)][AA]

is greater that W^ NE

The

acceleration rate terra can therefore not be eliminated if the


system is to
here,

remain stable. The equation that has been considered

if it is physically possible to mechanize it, appears to


It remains to be seen whether

provide a satisfactory system.

additional terms in the mechanization equation can improve the


response.
E.

Consider next, therefore, the oerformance which results

when to the

mechanization equation just considered, the first


is added to the second and third. This mech-

integral of A r a pp) (

anization can be written

JJ/Km ***+V*ob*j?Ki*

dt

* +s bwiow /*.* dk-t-StUKaaa A =JffK^


rlM)

dtdtdt

+5

tc , cm)]t ,

w ^/A ppl dt dt

+\m ^,u,Jk^ *

(IV - 19)

CONFIDENTIAL

from which,

with eq (IV-3),

Lc(cm)]ffAj

brUnM

(IV-20)

^Sind)|onq

'

\fcHlAAD

^M,'^) long

'

^[C(cmUC A A] (Dw(in<l)

Ug

Further substituting eqs (IV-1) and (IV-2) into eq (IV-20),

5 CC(cw13C'aa1 H, (trot) ,\ +
c

'[CfOYlflLAA]
_,,
i

NE

5lc<cw03 [AAH

>)-(tru.e.)

+
**ME
"^ccfcm^LA a]

NE

5 J [C(C^][AA]/

^/' V +^=^L(t)A + K(^T^K+ |^^(t)A ^


c-

(C)A r
[c(cw)][a'A]

^[cfcwBCA

AT

[c(cml]L"A A]

(IV-21)

Here again, the lowest order forced-error term contains the

acceleration of the wind. The coefficient of this term is now

and does not contain any controllable sensitivity. This

NE

system is therefore less satisfactory than that of par 2-D.


F.
its

For the sake of completeness, A r a pp) is next added to


(

integrals in the mechanization equation just discussed. This

gives

A Wmd) dtdtdt +S ^
cc(c

//A r6hd) [AA3

dt dt

+5

{c(

HAM

/ A,. 6nd

dt

fcfcwa ft'A]

A rfm<

i)'

A rapp) dtdtdt+5

[c(cm11CAA3

/ /Arcapp)

at dt +o[t

,,

cml]rXA]

/A ^ pp) dt -r o Cc(cm):i
r

CA

r(app)

(IV-22)

from which,

with eq (IV-3),
m

^fc(cwO][AA] ng fcrcmW[A A] ^tind)loq~'"S


'

C ^

5 tc(C*01CAAl (DQmd)Umq " '(undjlo^ ^^[cfc^KAA3^^Cma)Wg"T Wri^)l [cfcv

'

O
74

(1V-23)

CONFIDENTIAL

Further substituting eqs (IV-1) and (IV-2) into eq (IV-23)

^Vftrue)

NE

W* 'ME

5, C
&fcin)JLAA]

5
T 5
"vftrue)
\,/t 1

"Vtru)

\Ji e

A.*...., -ft rue)

=(C)A r

S mm xu
'

(
'rc(cm)][AW

\C)A r

^^

(c)K

+
5 [cfcn-OH AA]

fc)A,
(IV-24)

A]

This system, like that of

pa

r 2-E,

contains an unremovable

steady-state- error proportional to the magnitude of the wind

acceleration.
G.

When examining possible mechanization equations, it

next seems reasonable to add to the system of par 2-D the fourth

integral of A r (ind) us ^ ng tne fourth integral of A r app ). This


(

gives

A^dtdtdtdt + S,
from which, with (IV-3),

rcfcmyjtAA.]

A rM dtdtdt+5 [cfcmyCA JjK<\rt<k&+\tcrtn*jKtoi) *+V*)3tXiaA

rtI n

i,=

/WA ^
r

p)

dt dt dtdt

(IV-25)

^[C(cm)][AAl

^r(ind)

'

^CcfcmutAAl "r(ind)

^fc(cm\][A>]^Hind)

'

^[cfcmul'A"A] 'Mi'nd)

- ^
'

'Cindll

(IV-26)

Further substituting eqs (IV-1) and (IV-2) into (IV-26),

75

CONFIDENTS
5
is CcCcitJKaa]

WIS ^tCfcW
NE

r ftr

^V ft rue)
^OCa'a]

k^-g""

^ A^=(OA**+fas-a

'1

<)A,+^^

fC)A r

Cc(cm)] Caa3

(C)A r +

fOA
*u (ci-tOK'a'aJ

(IV-27)

CcCcm^rAA]

Cc(cwnCA~A3

As was true with the corresponding system of par 2-C, an


error caused by the velocity of the wind results.
H.

When the third integral of A r app


(

is added to the

fourth integral used in par 2-G, the equation becomes

A^, dt
from which,

dt dt dt

+ S K(cmaUA]

///A r(inJ) dtdt dt + S CCfcmtfCAA]

rffn j)

dtdt43 DC(tmac

A A1

//Vfinj)

dt+5

cc(cm , ]rXA3

r(lhJ)

7/L

pp)

dt dt d: dt

+ S [cfcrrOJO JJjA rh?p,<kdtdi


(IV-23)

with eq (IV-3)

^CdcmlKAAl A(md)~'~^Lc(cw>)D[AA3 ^rCind)

'WmncTA] ^V(md)

^^'fi'hd)lon4

'

^ccCarOKAA} VlAVfmd

(IV-29)

Further substituting eq (IV-1)

into eq (IV-29),

S
rcfcwXlfA'AT

'CCfcm-PCAAl

"
VV N
,-,

fcrtl)

Ccfcm1ilt'rAlJ

'cc^cmnfA-A]

ft

.= (C)A

+3

cfcw,3cXA]

fc)A r

-h -^aaa fc)^ +
Ccfcmqc'AA]

^^ ^ ^
'ccfchOH'X'A]

(c)
Ccfem!

Acfc m)3C*AA3

KA

(iv-30)

A3

Here, as with the system of par 2-D, the control equation


indicates that the lowest order of error in the steady state
results from the acceleration of the wind. The coefficient of

thiatenais

S tC(CM)][AA]

s[c(cm)][Xa]

v2

s[c(cm)Haa!
76

cc
As before,

if S [c(CM)l

[*AA]

is raad e equal to

the

w NE

acceleration term disappears. The coefficient of the acceleration


rate term is

MCM)UJAl _
S

LC(CM)]UA]
.

appears

[C(CM)]UA]

w 2 ne

[c(cm)K7a]
[C
(

that this terra can be made to disappear if


s

CM )]

AA

^s

[c(cm)][aa3

chosen equal to

J2 W

NE
a

From Routh's stability criteria for

quartic, the 3ign of

each terra in the homogeneous equation must be positive, and

O
^rc(cm):c*AAl

e
[c(cm1][AA]
^[cfcm)]

^[C(CW)]tAA]

LA

A]

pCcfcmKAAD

l^[cfcm13LA'*]J

>0

lv-31

If S [c(CM)][AAl is made

ecl ual

t0

W NE

eq

(IV-31) becomes

![cfcm)]:/> A]

TcCcmDCAAJ

'rc(Cfn)3CAA]

5 ^[c (cm) -'CcfcmUQA AD ^5


J

'[cfcm)][ a A]
3

[a A]

J[c(cmi]f A A]

>o
(iv-32)

ME

A A]

Therefore, in order to eliminate the acceleration rate term,


"

[C(CD)][Ta1

s [C(CU)1 Caa]]
*
(

mu3t be

grater than

zero, which

is possible only if S c CD [

j]

AA ] becomes negative. Routh's first

criterion of stability, which forbids variation in the signs of the


quartic coefficient, is then violated, and the system is unstable.

rr

ONFtDE;;,
Once again,
in

therefore, it has not been possible to eliminate,

the steady state, any effect of wind of higher order than

acceleration. Mechanization eq (IV-23) appears to give the


same accuracy,
of eq

in the steady state, that results from the use

(IV-15). In choosing between these equations, it is

necessary to compare dynamic response and ease of mechanization.


Before an examination is made of this problem, a study will be
made of other
I.

mechanization equations.

In an effort to eliminate the effect of acceleration

rate, the

second integral of A r app


(

is next added to the mech-

anization equation of par 2-G. This gives

JJ/fr,^
from which,

dt dt dt dt

\ a,^.JJfA

,,,

dt dt dt

+S M< ^ a , fjh, M1 dt

dt

+S Wlwr.,y\W) A dt +3 ow , A,,,., -fjJfK>

* dt ck

dt +S

Mn JffA,

llp ,

dt dt dt +

3^

CAA1

Km)

dt dt
(iv-33)

with eq (IV-3),

long

(iv-34)
Further substituting eqs (IV-1) and (IV-2)

into eq (IV-34)

) |_
1

^Cc(cm13CAA1

ICCfcmOtAAj

5
A, (trut)
*^N6 ^cc(c>tila'a:

"r(true)

'

s CC(c m )]c'AAl V

^ccicmac'A'A:

NE

s ^Cc(cm)][ A

A r(true
A}

=)A> |^^
^Cck^\]CAW

fc)

+f^^
'Cc(cm)3C'A"Xl

(c)A r

+ ^^ (OA
5. :cicm
a a]

fC)A,
^tctcmniAA}

(iv-35)

This control equation shows that, instead of improving the


response, the use of eq (IV-33) has resulted in an acceleration
error which it is no longer possible to eliminate.

78

CONFIDENTIAL
J.
A

final effort to remove the acceleration rate term is


)

made by adding the fifth integral of A r (a pp


the mechanization equation used in par 2-G.

and of Ardnd) to

This give3,

rr nd)
,

dt dt dt

dtdt-S^^Jjff^ dt dt dtdt+5 ^
cc

)n , A

w^A K

nd)

dt <*t<it+

S^^ffA^dt A+S^^A^A + S ukmmn K^=[[[[[k^A

<* dt dt dt

+S^

lk

J[f[A ran)

dtdtdtdt

from which, with eq

(IV-3),

^Cc(c^][XA3^i-<i><i)

Ccfc~UtAAl^r(,a^

'

*^Cc(cv,OtrA]

"KinaV^Wfc^tacXAl

"f(i<IJ

'

''Hind) long

""

^ Cc

(c

mrj c AA3

lV\-)(ini)Uli

(IV-37)

Further substituting eqs (IV-1) and (IV-2) into eq (IV-37),

-A, (W)

5 ^LCtQvlH A AJ

5 ^CcfCmHCAA]

w tcfc13LAA3

(irut)

** CcfcmOfaVo

(trul

,
<

Cc(cmOLAAT

nE

^Cc(cmUC"A"A]J

'ccCcviUTa]

"]

rftrttf)

= fc) A, +

******
^DifcmacTA]

(C)A t

**'Sn
^Cc(cn,nc"A"A]

(C)K +

WE

s CCfcmnC

^^
^CcfeTlt'X'u

(CK-f

^^
^OtftmaCTw

CC)A r

'

(QA r
*^[c(ewOn"X"A]

')/

llv-38)

The coefficient of the wind acceleration term of the forced


error now has the form
Ls

K(CK)][ AAJ

Ic(cm)][Ta]

w ne

[c(cm)][Ta;

}
,

To eliminate this,

SrcfcMlir ^"1

must be made equal to *

W
NE

as

before. In this equation, the wind acceleration term has the

coefficient
.

S s

tC(CM)][AA]

[C(CM)][AA l

If this is

[C(CN)][Ta]

W2 NE S [c(CM gf7A ]

to be eliminated,
S
_

EC

(CM)] -

must be made equal to

[C(CM)1 [AA]

W NE

79

^NFIDENTIA L
Using Routh's stability criteria for a quintic, the coefficients of the homogeneous equation must all be positive,
and

5 K(cm)]CAA]
.

tolltTAl
tc(cmlH"AA]

CcfcmmAA-]

CCfc^nC

A'Al

\c(cWCAA) ^CC(C^CAA]j
C

WlKT

^^^{V*,,^
t0

^c^CAA3

^mXAD Acta*,

A^

{\<cm? ] {S,cfcwatT4j^cefc,)]rAK]

>o
ic(c*m"'nj

(IV-39)

The system is stable, with proper choice of the other co-

efficients,^

S C [ (CM)]L A A]

is made

e<? ual

~rw
NE

eliminating

the acceleration term. If this substitution is made in eq (IV-39)


and if in addition the substitution
made,
tC(CM)][ AA]
m

tP(CM)][ AA]

W NE

is

ccfc^nc AA]
[CCc^DllA A]

c>v)3C'a"a]

Cc(cm)][AAT

C
[cfcm)3[ A A ]

w N
ME

,y NE

V
CcrcmQu'A'Al

>0
s s ^CcfcW)]TAj
(

w
This equation simplifies to

CcfcrnflC X" A]

IV-40)

ME

CcCc^DC'a'a]

tcfcw)Ji:AA]

cc(c^):["aa]

80

;:'i.

CcCcm^c'A

a]

Wcc(c**03[a"a]

^ CcCcm)3[AA] -S Cc Cc mil
.

A a]

WE

CcfcmOCAA]
[cfcnnOC
a'

>0
- S 5 ^IcCc^llL'A All WCcfcm)3CA
Al

(IV-41)

a!

.wWE

Since all of the sensitivities are positive, this equation


2

can be divided by

(w z

nr S [C(CM)][TAj'
S[c(cm)][7a]

provided that

only S

[q

CD

)]

["aA

*3

maintained smaller than

w NE

]
/

without changing the inequality sign. This yields

...

S [CfcmVJLAAj

'[cternVJCVA]

Ml

5,CcCcwl]

A A3

CC(cnnlK'A"A ]

-JccCc^iICaa]

^[clfrnfl r a" A]
J

J^

IV-42)

or,

simplifying,

-S

c(cm)l

A A3

>0
(IV-43:

Since

Sf;

CM )] [*Ta 1

must be kept positive in order to satisfy

the first Routh criterion, both criteria cannot simultaneously


be satisfied, and the system becomes unstable if an attempt is

made to eliminate the acceleration rate steady-state error.


18

Implementation of Equation s
A.

Four mechanization equations have now been discovered

that eliminate

steady-state error caused by the acceleration

of the wind. It has not been possible to eliminate any higher

order terms, but is seems unlikely that any steady-state error

resulting from the acceleration rate of the wind can have appreciable effect. The simplest of these equations was eliminated

because natural frequency could not be controlled. The three


usable equations are (IV-15),
(IV-28), and (IV-36).
In eq

(IV-1S), two of the three sensitivities can be controlled;


eq

in

(IV-23), three of four sensitivities can be controlled; and


(IV-3t>),

in eq

four of five sensitivities can be controlled. In

order to obtain the most satisfactory possible dynamic response,


the use of one of the more complicated eauations may be indicated,

but this choice of equations is subordinate to the considerations


of ease and accuracy of mechanization.
A

practicable method of

implementing eqs (IV-15) and (IV-2S) is now presented, and the


complexities compared.
B.

In mechanizing eq (IV-15),
( )

(DC)

n( j)

(long)

is u3ed

instead of A r app

as the output of the pendulous accelerometer,


1

as explained in par

of this chapter. The resulting eq (IV-16) in integrated form:

is restated here for convenience,

'ttCcm'OCA a]

dt r A.,,.,, V(m4) u

+ S, Art.M^/yhxtau.j ^tcCci^BtA
a]

dt dt dt

+s tetw

JdcU h

dt dt

(IV-44)

'J

Now, an imaginary auxiliary direction is established, as

shown in Fig IV-1. This direction is controlled, with respect


to the reference vertical, by the equation

A(au ofreO ^EcfiioOUtAA]

/ /

W(ind)

\*n$

& "^

V-45)

The controlled member is positioned from the reference vertical

through the use of the auxiliary direction by the equation

"Kind)

^cc(c**)fau<y)[AA]

//(D c )

(1

M<

nuj dt

dt-5 Zc(CM)(i ^ [ilK J A

r(

^ Ahiikt dt

iv-46)

These equations, when combined with the relation

(IV-4-7)

become

</</

///

uv-48)

This is identical with the desired eq (IV-44), with the

relationships among sensitivities that

o
^[cfc^)fatm):[AA'J

J fcfc)][AA]
(IV-49)

^CCfc^lA

A]

KM

CONFIDENTIAL

(in<4)

cv ind)
"

long

(aux) (ref

(end) (quk)

Wcrt) fapp)

(Vert), (ref) ref

(VertK n Cind)

(Dir) (ayK)

PIG.1Y-1 AUXILIARY DIRECTION FOR CUBIC MECHf\NIIAT\ON

EQUATION

84

^[cfovO^CAA}
Cc(ci^)(dUX)XAAl

IV-50)

[c(a [c fa

i*y)"]

Ta

a]

^[cCcwODL

a AT

(IV-5D
Equations (IV-45) and (IV-46) can be mechanized easily, as
shown in Fig IV-2. The mechanization of eq (IV-15) is therefore

feasible.
C.

Equation (IV-23) is now to be mechanized in a manner

similar to that used for eq (IV-15). The integrated form of the


equation, using (DC (i nd (long) instead of A r ( a pp), as before
) )

becones

...

A,-+5

[c (c **)]

a a3

hfl

-^dt

+S

m x>J Arc^dt

dt=S lc(C^y]LAA].

(t)0(mJ)lo , dt dt dt

(DC)(iM)U ^dt dt dt dt
(IV-52)

For this equation, a pair of auxiliary directions is established, as shown in Fig IV-3. These directions are controlled,
with respect to the reference vertical, by the equations

Arau>vr^&~"5 t:cfeux)aD[AAj// (DC;(ihd)lo

dt dt

liv-53)

CONFIDENTIAL

< Z

Q D H
o z o

o
z o o

!<

Of o UJ < * t Q u <
-2

N
Z <

UJ -J
CL

- o
^r

UJ
-*

< o < o o -I
CO

o c

O O
86

CONFIDENTIAL

ld)(OUX)Q

(DC) (Lnd)donq)
A(lnd)(Qux)b

(QuxKreDa
(aux)(ref)b

Vert happ)
(Vert) (Lnd)
(

(Vert) (ref)

r '(aux)a

Dlr >(QUX)b

FIG. IV-3.

AUXILIARY DIRECTIONS FOR QUARTIC OR QUINTIC MECHANIZATION EQUATION.

87

CO/VF/DENTIAL
(IV-54)

The controlled member is positioned from the reference vertical

through the use of the auxiliary directions by the equation

^Mind)

^CcfGM&ioOHAA] /(i*d)(***)&

^Ccfc^faux):

Caa]// A(md)ra.Ob ^t dt
(iv-55)

These equations, when combined with the relations

(iv-56)

(iv-57)

become

t-(iVd)

Sccfc^fewOHAAJ ^Ccfe^lfau^DCA AD

l\(mi) 'VdVid) /

ut "^"^ccfcmVauOKAAi

//A Kl M
-

a)

"*

S BAluft ^ CA 'fa^CcCeVku<)3LAX] //A^Qmd)!.^

ut dt

Clt

~H

[c (

aU)0b[AA]

O ^^^[aa] / / // l^MmdJItmg

"*

Clt

Ot

This is identical with the desired eq (IV-52) with the


relationships among the sensitivities that

^CcCiux)a3LAA]

t
'

(iv-59)

Wcmu C A A3

CONFIDENTIAL
I

5,
(

IV-60)

CcfcmnCAAT
5,[c/crnVWODCAAj"

CC"i)]L'AA]

(IV-61)

CCfcrnQCAA]

'KtcmKauxnCAAJ
[c(cwO]tAAl

(IV-62)

Equations (IV-53),
almost as
Fig IV-4.

(IV- 54), and (IV-55) can be mechanized

easily as were eqs (IV-45) and (IV-46), as shown in


The choice between the cubic and quartic mechanization

equations is largely,
of the

therefore, to be determined by a comparison

dynamic responses of the system when these equations are

used.

D.
is

The method of mechanizing the quintic equation,

(IV-36),

similar to that for the cubic and quartic equations just dis-

cussed.

In integrated form, using (DC) ind (i ong


)

instead of

*r(app),

this equation becomes,

-1W^-A-Aa Ar^d)

^ccfcmtfc-A-Ai

/A r6 d) dt

+S

ccfc m):it A

w //A r(lnd) dt dt ^5

(c

cX A0

/ /

/A r(md)

di dt dt

Wcw)3Caa1

foc) md)

dt dt dt

dt+//

/7(VC%^ dt dt
89

dt dt dt

(iv-63)

CONFIDENTIAL
on c

in

< z
H O z
D

o
u

<

z o o z o a p u < Iz Zo UJ
UJ -J
a.

< N
z <

u. u O UJ 5 <

IT e>

<

u o
-I (0

o
90

CONFIDENTIAL
< 2 a

O X o
o

D o z o <
I-

2 5 D O u 2

UJ
LU

O ^ <
N

<
a:

CD

<

o O
-J

CO
in

5
91

o 3

CONFIDEN]
The
57)

two auxiliary directions established by eqs are required to implement this equation.

(IV-53,

54,

56, and

The controlled member is positioned by the use of these


equations and the equation

(IV-64)
When these are combined,

the control equation becomes

<;^H-S cc(cwX4U , )3l^

% /A r(md )dt

S c(cm^aumAAi //A w
C

nd)

dt dt

+J

CcfcmKau *, 3CA V]

yy /A

r (

md)

dt dt dt

^pei&u^b]tA] ^[ctcmHaurtitA-^ / / / /(l-'^Ld)io- jdt dt


1

dt dt

+o

Cc(aL)>)aKAA:s

o [cfeW )ft U< )][ A A] /////(DQ.nd)


-

dtdtdtdtdt
(iv-65)

This is identical with the desired eq (IV-63). The mechanization of the equations from which (IV-65)
shown in Fig IV-5.

is derived is

This mechanization is, of course, more

complex than the systems previously discussed.

Although the

mechanization is entirely feasible, the quintic equation will


not be

considered further unless the dynamic responses of the

system using the cubic and quartic control equations show

themselves to be unsatisfactory.
I.

The Closed-loop Syste m

In order to examine the two mechanization equations between


which choice is still to be made,
it is necessary to examine the

response of the closed-loop longitudinal system to wind forcing

functions, using each of these mechanization equations. Response

equations for particular mechanizations have been derived. The

CONFIDENIIAL
general closed-loop equations are derived in this section.
The general mechanization equation can be written, for

this study, as

(iv-66)

The linear distance of the missile measured along the track,

from the initial reference point, can be written as

~~

//

MS-fciro-dt

ut

+V DE-M3rmitiA

|)t

+ ^cv,^d})

iv-67)

Also,

X
~Ar(trae)
(IV-68)

These relationships, with eqs (IV-1, 2 and 3) and the smallangle pendulum equation

^C

4ru01on 3

^ Vie

""

^ ^iR

(,V " 69)

combine to form the closed-loop diagram of Fig IV-6.

93

CONFIDENTIAL

<
2 <
ex:

cr

CO

<

CD

1 <
i

u
C3

o
k

9'

c J

<*-

<
z o < N
1

i.

a o O
i

uJ

X
Z>

UJ
I

</)

O
c

a:
c:

5
+>

< I o UJ 2
t

O u
UJ

cr

o z
CO

cn

o u
i

c J

>
tn

Q
a:

<
o z
o
1

CJ

a:

o z

CO

z g <

CO

u
a.

o
UJ

o o
/ i

z <
a:

<
-f-3

1 en

a. _J

a z
UJ a.
i
i

g
c

B
-

a
*

c
UJ

>
L

UJ

94

CONFIDENTIAL

The initial conditions in eq (IV-67) do not affect the


e

response of (C)A r to V

rj;

(air)]

and can ^e ta ^ en as zero.

Figure IV-6 can then be simplified to Fig IV-7.


In accordance with servomechanism theory, the overall

performance equation of the range indication system can be

written

rD c) r/ rris)

JcK =
vv

fr F

U4 frBLdo -fl*L)
I

\(pc)

(IV-70)

where (PF) (pcnd)

-i6

and (PF) (kin)

-iP RE

The

IR

mechanisation equations to be examined can be written

~l~

(PF)
Cc<ci*03Ca

5 CcfcwOC A A3 P

(IV-7D

"^Cc(c^)]LAA3

(PF)rmccKKqturtW

c
^Cc(cw,tlCAA3

h x -l-S ^fcfcmuftAl P r

r n*4- S
'

^Ccfc^)3CA'A3

ft

(IV-72)

Using these equations, numerical values can be assigned to


the sensitivities, eq (IV-69) solved for various types of wind

forcing functions, and the responses using eqs (IV-73) and


(IV-71) compared. Numerical values are first assigned to the
95

>

'

>

WWIDENTIAL
U

-V
v_^
4

< u

>

ar

CD

o
i
f

^ O

2
<

<
.v_>

c:

<
^^
i

L.

CL

E ^_v^
u_
v_^

Q_
i

a
UJ
a)

si C o
4J
r> i_

-a

>^

v-/

c
>

U.
1

u
or

_l

a
o z
ca

< u <i
i

UJ

z <
a: UJ

o
CJ
i

a|oi

^ v^f V
v_^

^^
<o
"3

CL

Q_

,,

s.

TO

UJ
.

CO
cc

<u

DC

Q. r^o
*"-,_,-*

UCL *"
~i
1
I

d)
ii

II

>B

o v^

CT J

c
Q.

u.

>

.iaJ.

96

CONFIDENTIAL
cubic mechanization eq (IV -*1), and then to the quartlc eq flV-72).
5.

Selection of Numerical Values

A. The homogeneous control equation for the cubic system can be factored into a first order and a quadratic term, becoming

fr+zrOfp^CDIflWp+W^O, CT
which expands
Into

(IV -73)

2(DR)IV+

(P)
Comparing
coefficients

P*+

2(DR)W
(ct;
J

P+ CT =0

(IV -74)

term by term, with

CCT)
Cc(cm)JC'A'AT

(IV-75)

CDR) =

W -W
2

(IV-76)

2W(CT)\A/:,

2(PR)
Cc(cm)][AA]

(CTJ+

(IV -77)

Then, for any desired characteristic time and quadratic natural frequency, a damping ratio and set of sensitivities can be computed.
Notice

97

v *< ivui\

that,

once the characteristic time and the natural frequency have been

selected, no control remains over the damping.

B.

The homogeneous control equation for

the quartic

system can be facsystem, these are

tored into two quadratic terms.

As

in the track control

made
2

equal, giving,

(p

+2(DR)Wp+W

)(p

+2(DR)Wp+W>0

w^*

This expands into

p +4(DR)W|D +

2W

[2(DR)+l]p

+4(DR)Wp+\AT=0

ov-w)

Equating coefficients with those of the quartic control equation, with


Sr^
[C (cm)]|A
/

virV"*A

* 2
NE

in

order to remove the acceleration term in the

forcing function,

^Ec(cm)UAAl

=
va/4

(IV-80)

(DR) =

'W

-2W Ne
(iv-*i)
NE

4W

Srctcm)]r ^

= 4(DR)

av-

W
38

CONFIDENTIAL

4(DR>

Plots follow for responses of the cubic and quartic control equations,

with various sensitivities.

93

CONFiDLM SAL
o
o

o
X
0-

CVJ

0)

O m

Z. UJ
(/)

n m n d - o u
it ii

S Q

o
<*

en uj *-

Q O o K ui
ifl

3 o * a
UJ
UJ

<

I-

Q_
..I

o
r<~>

o
o

5
O
<\l

<
(T

O Q
CD

O
_O
UJ
</)

o
Q.
(/)

UJ
cr

O
(

o
1

o to
1

o
i

o
<o
i

S3im "ivounvN
U0H3 01 IVnOB
100
)

ni

&
NI
J

39NVU Q31V0IQNI

NI

S3inNIIN

y(Q)

CONFIDENTIAL
z

< IO
cr
Q-

CM

O
LJ

(f)

>UJ
</)

Z o 2 < z - ^ Q
^

o o
(Z UJ
Q_

i < " o
<

o u_ o
UJ
c/)

z o
GL en UJ

o
i

o
CJ
1

o
1

o
1

o
iT)
I

CS3~iiw

nvounvN
)

ni
NI
J

B
V(0)
u.

39NVH Q3iV0iaNI

NI

H0UU3 01 !Vn03 S310NIW

101

C0NF1DFNTIAI
o
30

O < O Q
Z

o
(0

m < 2 z x
,

CVJ

tr o . Q H
or

o m

LlI

5 o 5 o Q 0. z < * Z _ o 00 2 z 3 O UJ H
c/>

c/>
Ixl

2
oro

z o
0)

/
1

O
CvJ

< a. 5 O o

i

cr

H
1
.

<D
[L

o
CUBI

s
UJ
SYST

cH ty
Z>>*
-

o
<*

o
1

o
i

o
JO
1

O O in
i

CS3im nvounvN
39NVW Q31V0IQNI
NI

ni
NI
J

U0U3 01 1V003) S3inNIW


102

\/(0)

CONFIDENTIAL

CHAPTER V
CONCLUSIONS AND SUGGESTIONS

1.

Disturbances to the Control Systems


In

examining

the

success of the mechanization systems whose responses


it

have been analyzed in the preceding chapters,

is

necessary

to consider the

types of disturbing inputs which will affect the systems.

These disturbances,

or winds, are of three major types.


fairly high frequency

First, there are gust disturbances of

and short duration. These are onormously attenuated


of the

by the long time constants

systems analyzed, and should cause no

trouble. Second, there are acceleration pulse type Inputs, such as would be

encountered when crossing weather fronts, where the wind velocity changes

very rapidly

to a finite

and fairly constant value. These can be analyzed by

the step function

responses of Chapters HI and IV. In Fig V-l, the path

through the "eye" of a hurricane has been approximated with step functions
of acceleration, and in Fig V-2, the

response of the track control system

is

shown

to be satisfactory,
is,

even with this violent disturbance.

There

however, a third type of disturbance; that in which a wind of

approximately constant velocity slowly rotates.* This condition has been

approximated by sinusoidal Inputs as shown in Chapters in and IV. The error


if

the sinusoidal forcing function continues for a long period of time, is

seen

to be

very large.

wind

of about forty knots, rotating with a

period of about

one hour, will introduce, in the steady state, deviations from the track, in
the track control

system, of about 20 miles.

*Ref. A. G. Bogosian Notebook dated

March
103

18, 1947.

CONFIDENT
"EYE" OF THE

HURRICANE

TRACK

VELOCITY I00MPH.

100

MP H.

ACCELERATION
2 2^250 FT/MIN

22,500 FT/Mlty

jr_

_j^0,Z50 FT/Mirt

FIG. V-l.

APPROXIMATION OF A HURRICANE USING ACCELERATION PULSES.


104

CONFIDENTIAL

S31IW IVOUnVM

Nl

X3VH1
1Q6

INOtiJ

NOJ1VIA30

CONFIDENTIAL
Three ways of correcting
a.

this condition suggest themselves:

Build systems with variable coefficients, which provide the neces-

sary high damping with step-inputs, and have a different response to such

low -acceleration inputs as sinusoidally varying winds.


b.

Increase the order of the track control system so that the forced

error results only from high order greatly attenuated terms.


c.

Modify the system

to

reduce the effects of the forced errors.


all

These three methods are

examined

in the next section of this chapter.

2.

Suggestions for Improvement of System Response

A.

Track Control
a.

variable coefficient system can be of value only


of the

if

changing the

damping and period

system

will

improve the response. Since the co-

efficient of the acceleration rate forced error

term increases rapidly with

increase in damping,

it

appears that this should be possible.

series of

track control responses using the

same magnitude

of disturbance but difIt

ferent damping ratios has therefore been examined, Fig. V-3.


that no choice of
the forced

is

seen

damping

is satisfactory.

If

damping

is entirely

removed,

error of the simplified system reduces


still

to zero.

Actually, a
if

forced error will

exist

if

there is any vertical wind acceleration, or

non-ideal components are used.


is equal to that of the

Then,

if

the frequency of the rotating wind

system, very large errors can result. This approach,

therefore, does not provide the needed improvement of response.


b.

The purpose

of this section is to investigate the

response of a

quintic

performance equation. From eq m-26


written:

the

performance equation

may be

106

CONFIDENTIAL

(Tm) (DR)

PERIOD OF ROTATING WIND

A (Vert)(tc)(max) N MINUTES OF ARC ? [EQUAL TO NAUTICAL MILES OFF TRACK]


7.735 14.310

1.875 1.875 1.500

10
15 15

13.620
7.600
13.033

1.050 1.875
1.875

15

20 30 30

9.548
14.899
13.652

1.500
1.050

30
40
40

1.500

12.080
14.096

1.050

MAXIMUM TRACK ERROR FOR VARIOUS OF 31.4 MPH

(DR)s

WITH ROTATING WINDS

FIGURE V-3

1Q7

CONFIDENTIAL

w NE
(C)A(Vert) (be)

v-i
LA'A-3

_
p'+i
giR

S[C(CD)]

^t(a r)-M]
L

V [E-(a L r)J

+
* [(oir)-M]

'[C(CD)]LA Al
'[C(CD] [A A]

p%
i

w; E+
-

>a<

S [C(CD)] [AM
[C(CD)J

'ECfCDJDCA'Al
.

'[C(CD)H A A]

[A AJ

^ECCCD)J[AA]

P^ "NE

va/2

S[ C tCD)]LA A]
7T"
tC(CD)]

TCCcD)3tA A]

LA A]
J

P +

w;

J [C(CD)][A Aj
'[CCCD)][AA]

CONFIDENT!,

For
equation

this study it is

assumed

that the

denominator

of the

performance

may

be factored into the form:

(p

^KpS

2(0R)W p+ W*)

Let

Wo

KWuE

where

W NE A

Re

Then expanding and equating coefficients

of like items,

i//rxD\\A/

9lR
->LCCCD)][A A]v[( aLv .)-M]

_.

S lC(CD)][A A)

v-2

CT

D [C(CO)][AA]

2K

W NE +4K(DR) W NE
(DR)-WNE

+^
+ 4K (DR)W
3
-

W NE
=w
.

'5^^
+
Sr

2K-W-NE ^4K 2 CT

5 K c)CCD)3tAAt
S[(C)(C0)3C'A "A!

a c D )3CAA]

v^4

J[C(CO)][AA]

4K

(0R)Wne

K4 W 4

- W\,

5 CCCCO)][AA]
3[C(CD)][AA]

Y-5

CT

601

CONFIDENTIAL

W NE
CT

^[CCCD)][A A]

S [C(CD)][AA]
3, 4, 5,

(V " 6)

Examination of eqs (V-2,


selected, then

and

6)

reveals that

if

and (DR) are

CT

and

all the sensitivities

are fixed. General expressions

are now developed for determining the numerical values of


ratios of sensitivities.

CT

& ad ui certain

From

simultaneous solution of eqs (V-3 and

5)

CT

4K(DR)(K 2 -1) W NE (-K 4 +6K*-1)

(V " 7)

From eq

(V-6)

5 CcCcd)J[aa]
5[cCco)]rAA]

K3
~

W* E (-K 4 +6K*-1)
4(DR)(K 2 -1)
(v " 8)

From eqs

(V _ 5 and 7),

S Lc(cd)][aa]
5[CCCD)][AA]

K
=

W^ E (5K
K 2 -l

-1)
CV-9)

From eqs

(V-4, 6 and 7),

S [C(CE))][AA]

T~?

^[CCC0)3CA A]

wNE

2 K +4 K [3(DR) - 2]+K [l3+6(PR) ]-aK[l+2(PR)*]


7

(V-10)

4(DR)(K 2 -1)

uo

CONFlDt
From eqs
(V~2, 4 and 6)

M
2

9ir
'CCtCD)]['A

_ |-K + 4K C2-3CDR)
fe

J+2K [-7+4(DR) ]~4K'C2-3(DR) 3-l


4K(DR)(K
of
2

V_li
1

AJ V [(Qir)-M]

-1)

Limited time permits the evaluation of only one numerical solution


the

performance equation.
Let

= 2 and(D*)= 1.

Then,

5 rrrr mi ta/a "[C(CD)][AA]


T

i4 x^

l/v. V

NE

S[C(CD)][A A]

S[C(.CD)HA A]
5[C(CD)]CA'A]

76W NE

^[C(CD)DCAA]
'[CtCD)][A A]

__

103

WNE

<3iR
5[C(CD)][AA]V[(Ql r)-IVl]

625
24

w NE

CT
input Vr
_
/

24

w NE
2

Substituting these values into the performance equation and using as an

ai

vi

= 2000 ft/min

sin

t,

III

CONFIDENTIAL
(C)

(Vert) (tc)
this

.8

minutes of arc

(.8 nautical

miles)

Comparing

answer

to those obtained in the plot of results in

Chap-

ter in indicates that the quintic performance equation produces a better re-

sponse than the quartic one.

It

should be mentioned that no effort has been


for the quintic.

made

to select
It is

optimum parameters

c.

seen, by an examination of the track control system perform-

ance function, that no control over the gain setting exists. As was shown in

Chapter in,

it

is desirable to

have the sensitivities negative for the quartic


this change,

performance equation. Making

eq in-25 becomes;

w
P
(PF) (tc)(CL)[VA]

=
4

TC(CD)1CAA] V C(cur)-M]
teCCQflCAAl

V-12
3

P
where
If

5 [C(CD)] [A

9lR
^ [C(CD)][A A] V [(air) -M] P

5[C(CD)][AA}
'[CCCD)KA A3

va/2

_2

S [C(CD)][AAl
[CtCD)][AA]

va/2

S[CCCQ)1CAA]
'[CCCO^tAA]

XkJ

A]

sensitivities are

now

positive numbers.

a controllable gain coefficient could be introduced into the numerator,

so that an attenuation of the forced error could be provided, then the system

might be made to perform without excessive errors. This could be done,


appears,
if

it

the

wind acceleration term could be attenuated after entering the

system. This cannot be done, but an approximation of this desired result


is

achieved

if

additional feedback loops are provided as indicated in figure


(

V-4. The performance equation for this system, when

PF )( aero )(i nH)

is

assumed

to

be exactly equal to (PF)/

aero \|AA 1
*]

_ {[(Pfyaax)CAAl
1

""

^\?u) [V A] ^ P ^(Qro)CAA] ^ P ^ (mech)UA]+ ^Cpu)CVAl} I'^GLAA]" *}


[(PF^ aujf)CAA] -l
"( PF

V-I3

+ CPF) (aero

CAA] (PF) (mecK)[AA1

\ ai4x)[AA]^ Pf: ^GrAA]]

SII

CONF/DENT.W!

v, (E-(air)]

V[E-(0 Lr)J

IR

(DC) (truextc)

COMPARATOR

(DC) (Lnd)ttc)

(CM (vertxtc)
COMPARATOR

-V[,(QtD-M]
k

ZCcm)

*H(cm)

l~g

ME-(Qtr)]

Mz(cm)

WW)(AA]
IR

(wA&srtxtc)

t
COMPARATOR

-(air)]

ux#A]

9w"()rtXt0
m

g. MR

'

(OA w,r Wt)(tc)

V[(ain-M]
'IR

'Zccrn)

(PF) (aero)(ind)[AA]
[(QlD-M]
k

2(cm)

MR

(PF) ( pu)[vA]

gm

(PF) (aero)[AA]
.i2

_"V[(Q(H-M] Q

g |R

(P>

"*"

)(Q UX )[AA]

5(aux)[AA]

^(aux)[AA]

p%w NE
(PF) G[AA]

(ton
ur/(aero)ancj)[AA]

j V[CQir)-M](md) n
g
r

(.PF)(meCh)

= MECHANIZATION

FROM CHAPTER THREE

FIG. V-4.

MODIFIED

TRACK CONTROL SYSTEM.


en

UWIDtAu ,al
hen the derivative auxiliary feedback sensitivity is set at zero, this becomes,
ith

proportional plus first and second integral mechanization equation,

I
F^'itc)(CL)[vA-

^e
e

P-b
9lR
C

L^lc(cd)Haa] \

s Re
\

/Re
1 ,

+
[

5[CCC0)][A^]
\ScC(COj]LAA]V[(oir)-*IJ

f5( QlAX )[

i)
M- 14
...2

'LC(CD)KAA]
M/Z
|
,

tCCCDlKAA]

D [C(CD)J[AA]
In this equation,
Into the

^rC(CDH[AA] V C(q(.v-1-M]

P
/

+1=

S CC(COl][AA]

^ e 3(aiu)CAA] W NE lp +0(au*)LAA]
<-

5 CC(CD>HAA]

W NE

.p+0((W)LAA]

S [c(C0 ,]QftA] _1
^LCICOUCAA]

W NE

...?

W[c(CD|][.a

tfl

*>[C(CD)][AA]

a certain amount of gain control has been introduced


of introducing a

numerator, at the expense

term giving a forced

error caused by wind acceleration.

Routh's stability criteria show that this


Limitations of
of this control
i

system, with proper selection of sensitivities, is stable.


time did not permit the authors to

make a complete analysis


random

system.

However, a sample

result, selected at

is calculated.

The

denominator is first factored into two unequal quadratics, and the coefficients
squared.

They become:

/ S[C(CO)J[AA]
I

\
/

W Wa
t

V-15

5[c(CD)KA Al

5 (aax )[ A A]WK, E
,

Plc(cd)][aa]

iScC(CD)3C A A]

V
\

rw. -j-w. +?mQ\.(riQ)_ WW, W W2 -[VV+VV+ZCDR^DR), VotAAiWif ^


.

_vu

V-16

cc(co,]tAAi\
5[CCCD)][A'A3/

^CDRhW 1 W^ -fCDR) 2
^

W W2
^
1

SfamOCAA]

W NE
/

CCCCD)]CAA]

V CCatr)-M]

\SlC(C0)3C A A]

114

lAjmnucii IML
i

The values chosen were


S

(aux)[AA]
1

0A
1.5W

W
W2
(DR)
2

NE

'

W NE
0.7

These gave, using the equations given above, (DR).

9.18,

[CIC " lflfll

!
rccco)]CAA]

0.

124 , (f

"=">"1
)

= 22.64 and

IR

20.72

'[C(CD)][A(V]

SrciCUKAA]^t(Oif)-Ml

The magnitude of the


miles,

maximum

forced error was found to be 28.5 nautical

and the disturbance which gave this

maximum

deviation

was

a sinu-

soidally

varying wind with a velocity magnitude of 265.5 nautical miles per

hour.

This is a considerable improvement over the uncompensated system.


\f*

Next, S/
(aux)[AA]

was se * a * umtv

remove

the accleration term, and

was

set, at a

random value.

The performance equation, with quartic mechanization, becomes

'(aux)CAAD

5
(PF)
(l

'(auvc)[AA]

'CCCCDiJCAA]

z_

'CCCCOIICAA]

'CCCCDJKA A
;

9m
R E S CC(CDJJCAA
c
'[C(C0)J[AA]
I

18

5[C(C0)J[AA]

[C(CD)JLA AJ
CC(CD)][AA]

VfJfcurJ-M]

cKCULVA]

wcccconcAA:)

5 (ou*)cAA3

W ne

9lR
'[(airl-MJ^tCCCDUCAAj

5 (auK)LAA] \N NE 4-WNE p

+1^

^5 (QU)< )LAAlWNt i- r

...2

Scc(cO)HAA]

CCOCDrtCAA]

^CtCOUt A A]

D rC(C0)7fAA]

W
.

'cccconcAA]

115

CONFIDENTIAL
Equating coefficients, as before,
to those of a quartic with the

desired

characteristics, and selecting, at random, the following values:


S

(aux)[AA]
=
l

20

W2

2W NE

"1 ~ .0885 Answers were obtained making r^ TGCCOnCAA]

(OR),

*(0R) 2 = 2.75

(" CCCCP)]CAA]
,

= 478 andf

9lR
[CCCD)][AA] [(ai-)-M],
v

\=3.25

With the same wind input as

that used in the preceding


is

example, the maxian improvement

mum

error was found to be 48.8 miles. This, again,

over the uncompensated system.


It

appears, since in the first example the error was caused chiefly by

the acceleration
of

term, and

in the

second example by the second time rate


combination of the two systems, varying

change
(

of acceleration, that a

both s

aux )[ AA

and S/

aux xr^ A

might give a satisfactory solution. Time


simulator would greatly facilitate the work

prevents such a study here.


of selecting

optimum

sensitivities.

B.

Range Indication

Lack

of

time prevented an elaborate examination of the response of

the range indication


trol

system

to large

ranges of inputs. Since the track con-

system indicates

that the critical response is to sinusoidal inputs, a

series of solutions was made for cubic and quartic responses to a sinusoidal acceleration input with a period of thirty minutes.

As indicated in the plots of Chapter IV,


damping ratio of

the

response

is

improved as

the

the quadratic in the cubic equation, or the two equal quad-

ratics in the quartic equation, is decreased.

At a damping ratio near

0.3,

116

CONFi.
3.

HAL
TRACK CONTROL MECHANISM SUMMARY
PERFORMANCE EQUATION AMD REMARKS

MECHANIZATION

EQUATION

R E w' NE
(C)A (Vert)(tc)
*

<-

vy

AzCCm)

S[c(C0>]

[AA](DC) tl dHtc)

[c(CD)][AA] V [ia,r)-M]

^^(Vert)ac) *

W N e(U A

w: ME
CV e rt)(tc)

<?

[E- (air)]

[C(CD)][AA] V [cair)-M]

velocity

of

Wind introduces forced error

'^-

'"(Vert) (tc)

w,ME.

lL)A
'IR

^ rt)(tc) -

w WE
Vr,

Af

cm)

" S [c(cD)][AA]y f DC )cHto dt

-n
'[c(CD)][AA]

ir ^[c(CO)][AA]V[la>Y) . M]

THIS

SVSTEM HAS NO DAMPING AND BECOMES

Ki

5TA6 L E

IF

ATTEMPT

IS

MADE

TO ELIMINATE

FORCED ERROR INTRODUCED BY

A CC ELERPvTvON

OF ThE VMIMO.

[C(CD)][AA]
"*

[<QiO-M]

S d

[C(CD)][AA]

.V T [fa>0-M]
<C)A,(VertXtc) v<rt

'^'"(Vert)(tO

(C)A
'ie

- S A M [C(CD)]rAA]/7(DC ^od)(U)^ dt 2(Cm)


THIS.

R_

[E-(a;^,] ME-(a'.rV

SYSTEM

IS

UNSTABLE BECAUSE THE FIRST ORDER "TERM

vs

MlSS\KJ&

HII


CONFIDENTIAL
MECHANIZATION EQUATION
P.

TRACK CONTROL MECHANISM SUMMARY (.COWT) PERFOEMANCE EQUATIOV4 ^MO EtMAtK-S

w"
^'^(Vtr+)(*0

(C)A

+W

i+
[cico)][AA]
[(q.--)-m:
.

v,[E-(o'.r)]

J^i(C^)^

'

^[C(CO)][AA]

^^(md) (vc)

THIS

SV5TEM

IS

VJNSTA&l-E

bECAUSE CONSTANT

TERM

IS

MISSING

(C)A (Vfrt)(tc) + W VWH "me


1

__
ly
[lqir)-M]

[c(co>][AA]

(C)A

"KCn*" ^[C(CD)]UA]^Rind)(tc)"^lp(CO)l[AA]y^C)(in<))(tc)

dt

^y

CVeYt) (tc^

NE

(C)A

CVYt) Ctc

/W

S[c(.coV][AA]

...
VV-'

WE

H (VtA)(tc-)

_
C

w'
\l
,

[E-(a>'l ]
-j

>[CLCO)][AA]

[c(cO)l[AA]Vr_Ca r ,. M

ACCELtKATION

OF

THE WIND

INTRODUCE!. A FORCED

ERR.OR.

'IB.

5 [c(coi)[AA]

tt)Att ClJ(fc

*<

J [c(co
)]

(OL<+\W[ Nl Werl)(tO
'

*[c(cofl[AA]

(C)A

"5[Ctf.0fl[AA]

(ha
J[CCCD)][AA]

[C(cJ>^][aX]

**

we

5
[c(co)][AA]

OA

=
3 [clcoj]

-V
[AAjV^-^J

[aa]

V"' -H
1

c <c

D )] tA *J

[c(CD)]l.AA]

Stccco^AA^^UdHtc)

<*t

^
'IR.

KATE

OF

CHANGE

OF

ACCELEgATlON OF THE WIND INTRODUCED

A FORCED

ERR-OR.

W*
'[CtCDflCAAJ
[c(cd)][AAJ

[t(d][AA]

5[c<co)][aa]

[c(.cm][aa]

(OK ^t)(to)
^l (cmf S[ciCO )] [AA]'"Qm.>lt)" 5[c(C0)] [A 'i]J^\ii ) (tcf^
"

(C)A

+
S[c(CO)][AAi|

(C)A,., x*o +

yv

J [C(C U gtAA] V"-m]

LcUO\][AA]

Ml P rt(cc [cfcM][AAJ

(cu t.O(tc)
^
[c (

Nt
j

co)][AA]

[c(U][aa]

s[c(co)J[aa]

^''V.OtM

J [c(ti>g(A]*[c.<-)-M]

fr-Cairl]

\^m/M^M h^fffw*Jk* *

THE RATE OP RATE OF CHA.MGEOF WIND ACCELEEATlON PRODUCES A FORCED ERROR.

U9

CONFIDc
4.

Range Indication Mechanization Summary


Examining
the control equation obtained In Chapter IV,
it

is
to

seen that
and in-

it

is possible to eliminate forcing function

terms

of

orders up

cluding wind acceleration terms.

In a stable system, higher order


all

terms

cannot be removed.

general performance function including

mecha-

nization equations that eliminate acceleration and lower order terms can

be written

ine)

S(n-1)T

p + 5 (n)T

WcM(AA,(ion S )= p" + 5 (1)T p"-' +

-.+ S(n _ Z)T p*

Or-ra

If

more terms are included


if

in the

denominator, velocity and position


in the

errors result;

more teTms are included

numerator, control
If

is lost

over low order terms, while higher order terms are eliminated.
are omitted, instability results.

terms

block diagram showing a method of mechanizing the performance

function eq (vi9) follows, fig (V-5).

For both track control and range indication systems,

the

response for

sinusoidal forcing functions of 84.4 minute period appears to be the

same
-

regardless of the choice of damping or order of the equation in the

mechanization system.

120

CONFIDENTIAL

lS,KDC) (m

j,

lS q

f%-(DC) (inJ

;s

combinor

)[(f-+i)iDCUr

(Sa)

$2
p

ll^OlDCW
v

(S,l

tf--n)^DCUcl)r~S WW

AU

ind,.

sn

combtnor

CS 3 ) (DOiind),

Q (n-i)T
(n-i)

S w cnn
.

^(n-i)Tp (n-l)Tp

OiniT ^>
n -' p

(PF

(mech)lAA)

(loncp

Sci>T

5u)t + rz
,

^...j. ^'"^ Stn-Ejr


D (n-2)

^+S m

+ ---+S {n -in?

WRITTEN TO OBTAIN THE SENSITIVITIES OF THE EQUATION EXAMPLE: (*> WITH PROPER TERM OF BLOCK DIAGRAM

ABOVE, EQUATE COEFFICIENTS

uadra*ic)

S,
/\r(ind)
\0<]'

ft" DC) (m
) (

/S

<

>

so

S(h) T

S 2 S,

anci 5(n-i)T

~>

IMPLEMENTATION

OF GENERAL

S(ib)A roru

])

tS q )[-# pt
,

/vu-'v-'/iincijr (f+'HDCU,, p
v

S' S 2~3
p

'3

/\r(ind)_
"~

(s,)
combinor

'(f-|)(DCU
ArOnd)

-(#^

+ S c% )A

t(ini

S(n-n)
P

combinor
Ji

C On
P

r (

m J)

(S q
S(fb);

S(-fb) ?

S(fk)(n-2)

TION

OF GENERAL LONGITUDINAL MECHANIZATION

EQUATION

FIG.V-5
ISI

CONFJDEN

APPENDIX A
DISCUSSION OF SIMPLIFYING ASSUMPTIONS

Several simplifying assumptions were made

in

reaching the mechani-

zation equations and performance equations discussed and analyzed in this


thesis.
1.

These will now be examined. Omission


of the

Geodesic Acceleration

Term
the fact that navigation is

The geodesic acceleration arises from


performed on the surface
of the earth,

where

the gravity equipotential


to

forms the geoid. This geoid may be considered


roughness.

be an ellipsoid with

The deflection

of the vertical

due to this roughness never exis usually

ceeds seventy seconds


far smaller.

of arc,

and (especially over land masses)

While the missile travels over this ellipsoid, computation assumes


that the surface is a sphere.
this

The shortest distance between two points on


fall in this

sphere

is

a great circle. The gravity verticals

great circle

plane.

The great circle plane transfers

to the surface of the ellipsoid as


lie in this plane,

a plane section, but the gravity verticals no longer

but in

an arc intersecting
sile, traveling

it

at the points of departure

and destination. The mis-

along this curved path, experiences a horizontal acceleraof

tion.

The radius

curvature of this path, which is a

minimum

at 45

degrees

north or south latitude, is never less than one million miles.

The geodesic

acceleration is therefore of negligible magnitude for missiles with velocities of a

few thousand miles an hour.

l!22

CONFiDENTiAi

APPENDIX A
I

DISCUSSION OF SIMPLIFYING ASSUMPTIONS

Several simplifying assumptions were made in reaching the mechanization equations and performance equations discussed and analyzed in this
thesis.
1.

These will now be examined. Omission


of the

Geodesic Acceleration

Term
fact that navigation is

The geodesic acceleration arises from the


performed on the surface
of the earth,

where the gravity equipotential


to

forms the geoid. This geoid may be considered


roughness.

be an ellipsoid with

The deflection

of the vertical

due to this roughness never ex-

ceeds seventy seconds of arc, and (especially over land masses) is usually
far smaller.

While the missile travels over this ellipsoid, computation assumes


that the surface is a sphere.
this

The shortest distance between two points on


fall in this

sphere

is

a great circle. The gravity verticals

great circle

plane.

The great circle plane transfers

to the surface of the ellipsoid as


lie in this plane,

a plane section, but the gravity verticals no longer an arc intersecting


sile,
it

but in

at the points of departure

and destination. The mis-

traveling along this curved path, experiences a horizontal accelera-

tion.

The radius

of

curvature of this path, which is a

minimum

at 45

degrees

north or south latitude, is never less than one million miles. The geodesic
acceleration is therefore of negligible magnitude for missiles with velocities of a

few thousand miles an hour.

l!22

COJVF/DENTIA.
Two
figures (Fig A-l and 2) prepared by Dr.

W. Wrigley

of the

M.I.T. Instrumentation Laboratory, illustrate this geodesic acceleration.


2.

Compensation for Coriolls Acceleration

The acceleration
sile is

of the Coriolls arises

from

the fact that the

mis-

moving

in earth space,

which possesses an angular velocity with

respect to inertial space. This is demonstrated in the accompanying fig-

ure (Fig

A -3). Compensation

is independent of the

heading of the missile,

depending only on a knowledge of the earth's angular velocity with respect


to inertial space,

which is accurately known, on latitude and on ground

speed.

An

indication of latitude and of ground speed are available within

the missile, the accuracy of which are dependent on the success of the control system.

Errors resulting from inaccuracies

in these

measurements

will be negligible in any practicable system.

certain amount of cross -coupling of the track control and range


it

systems will occur through the Coriolls Computer, but


if

will

remain small

the component of missile velocity perpendicular to the track is not large.


3.

Small Angle Pendulum Assumption

For simplicity

of calculation, a linearization

assumption has been

made

in the output of the

pendulum. The pendulous accelerometer actually

solves the equation

[E

(cm)](hor)

^
_

(DC )

em
The control equations are based on
a
the assumption that

[E

(cm)](hor)

^^v

123

CO ilhDc
at

iAI

!
_ o

8 o i

i
ui

?
.

o o
Ol

bl

8
X
tfj

-i* I
UJ

o
UJ

a o UJ
UJ

or
i

UJ

O Z^<
tt

or
UJ

<OH

tt o n; uj

"

>
II

rf tu ** =r

z
(A)

Ul
UJ

UJ

* ar-" * uj < Zt
M

111

-J at

O UJ

O o w

O UJ

y P a
UJ

UJ

K3

a uj z

J
-j

uj

t
~E
f

oc

"o

O Sou
^UJ

OO

UJ

< o

o
</>

o: u.

J o i-

2
OC

< Ho
j
Ul
U.

o
if

OU

flC

*^ 3 ZOZ
ZUi

o
a
ul

2 o ul O o

y^imiaC

CONFIDENTIAL

125

UJiNNUtiNIIAL

<o

CO UJ
I

c
CD Ut

X u I- O QC < < Ol
UJ
UJ
tf>

u I
u < I UJ ui > UJ I o II-J

2 a: O u

2
Q
UJ I- QC

UJ

M
CVJ

UJ

X
u.

LJ

cr

I* OJ
II

>-

o O h O -J o UJ Ui > a
< a z O
CO ui

o o -J UJ UJ a > <o
QC

UJ

o h < a

II

v>

J O

<O uU o UJ
co

QC

UJ
-j

oc

< X UJ Z IZi*

O
O <

UJ

o u

oc

IO

<
U.

<o

Jr

<

oOz
=,fco

* CC

h- UJ

o< UJ O UJ
"IE
rqoTjv

?u

r
1

/.
.14
.

<I CD

~
3**

126
^

CONFIDENTIAL
The
validity of this

assumption has been tested, by comparing the out-

put, for the

track control system, to a pulse of wind acceleration of 2700


lasting for one minute, using the transcendental and the simpli-

feet/min

fied equation.
If

The results are shown

in the

accompanying figure (Fig A -4).

necessary, a tangent pick-off can be used on the accelerometer in-

stead of the linear pick-off postulated in this discussion.


4.

Neglect of Vertical Acceleration


If

the missile does not maintain absolutely constant altitude (or

if

the magnitude of the gravity force varies), the

pendulum angle will not be

a function of horizontal acceleration alone.


effect of an

This effect, together with the


lag, is

aerodynamic lag and a pendulum

considered in the folin

lowing treatment of the range indication problem.

There are changes

some

of the

geometric equations, as indicated:

Aerodynamic lag term:

QtE-tCmvKhor)

(A-l)

Vt(Q ir>-E]

(CT"Wro) P

+l

Accelerometer lag term:

(DC) (tru^
(

(A-2)

DO(trueKcorr)

(CT)(p U )P+l

A r(^,

a ie-^)Kho.->

(A . 3)

1?7

CONFIDENTIAL

D
Cl
.

OL
1

5 O z
tu
a>

>

Q.

^1

^7 ^ St
CO

oo

o z
KJ

ui

<
UJ

IS

(0

2g
o
UJ

o
u
LU
u. i-

UJ

< 6

<M
1

CO
1

<
1

o
i

s0
i

N
i

CO
1

<y>

(S31IIAJ

lVOULOVN) 38V HV"in3ai2 JO S31HNIW


128

u\^(^J9^)\f{o)

CONFIDENTIAL

(DCfcapp)

CDC)( Lnd)
(DC)(^ rue
)

(DC)( true )( COrr

^ r (opp)

^ r (opp)(corr)

& r (Lnd)
(C)A (Vert
)

WerO toP p)
(Vert) (Qpp) COrr
(
)

Wert)/ re f)

Wert)(lnd)
CVert) (.true)

FIG.

A-5.

INTRODUCTION OF PENDULUM LAG IN RANGE SYSTEM.


129

CONFIDENTIAL
lLE-U.mil

(DC) CtrueKCOrr)

q jR

a LE

(cm):(Ve rt)

(A-4)

The mechanization chosen was the cubic mechanization equation discussed


in

Chapter IV.

The control equation for the range indication system, using


cepts outlined above,

the con-

becomes,

(9lR~

Q-LE-CcmlKVert) ) //NTX

(C)A (.Vert
VlE-(air)}

p
_

^ U (pu)

("-

..

LlClcm)](AA)

y 1H

u (.E-Ccm)KVert)

<5
"

Lc(cm)]LAA1

(C T)(pq)

Q IR

S lcCcm)U AA1 Q IEp

cm)l(Vert)

(C T) (pu

]
1

r T

LcCOXH AA)
p+
l]

...

S tctcni)KAAl 9lR p

S(XCCm (AAUE-CcmiXVert)

(a-5)
2 3+ S[ C (c m> KAA] p [p

+ S Lc (cml]tAAl p + S [c(cm)U A A1 Hqi R -a

E-(cnKVItrt)ll(CT) p U)
(

"*

l(CT) (aero >

If

the sensitivities are chosen so that the cubic in the denominator beof a quadratic with (DR) = 0.715

comes the product

and a period

of forty

minutes, and a first order term with (CT) = 14 min, eq (A-5)

becomes

[E-(cmi](Vern

IC)A(.Vert)

-q^
(qi R -a

+L4W (^
f
,

( ,

Q [E-(cnQ3(VeTfl
9lR
3

HC

(rr)i 1

V[E-rt

[E

(verti)[4i;i P

V> q^ P^ ^E

[E- (cm)](Vet^

p 3P

CT )(pa) 4 W, E I 1
/
.

[E-(cm)](Vert) \
j

q IR

?
(A-6)

[(

CT Wo)P^][(CT) (p a)P + i]

130

CONFIDENTIAL
The
sile

effect of vertical acceleration


lift

becomes very

clear.

If

the

mis-

momentarily loses

and "drops", the denominator goes

to zero.

The numerator, meanwhile, acquires a term which gives a forced error

from wind acceleration. This error

will change sign with

change

in the

direction of vertical acceleration, but the effect of the denominator will

keep the opposing effects from balancing. The damping in the denominator
cubic, however, will return the missile to the track after the vertical ac-

celerations have passed.

Even with a downward acceleration greater than the


gravity, the

effect of

system retains

stability.
(

Reasonable values for

CT )( aero

and

^ T ^(du) w*^ vary


will, therefore,

rom

fraction of a second to a few seconds.


effect on the system.

These

have

little

5.

Assumption

of Perfect

Aerodynamic Response
that neglect of

Paragraph 4 indicates
can have
little effect

aerodynamic response terms


system. In

on the performance

of the longitudinal

the track control system, however, with a fourth order equation or higher,
the
left

system appears stable

if

a positive pendulum angle causes either a


It

or a right rudder deflection.

seems, therefore, that even a small


to

aerodynamic lag might cause the system

become

unstable.

The problem

was therefore recalculated for one choice


aerodynamic response, a pair

of (DR), using, instead of perfect

of equations.

The

first gives the desired

heading of the missile as a function of pendulum angle; the second relates


this desired heading to the actual heading of the missile through

an aero-

dynamic

lag.

131

CONFIDENTIAL

Az(cmxind) ~ S[c(cd)][aa] (DC)( in( n<tcr S[c(co)][AA]y (DC)(ind)(tc)(dt) ~S[c(cd)][AA]

J J (DC)(ind)(tc)dt dt

(A-?)

The variation system discussed


in

of the

response

of this

system from the idealized


in the plots

Chapters IE and IV could not be detected

drawn by
6.

the Differential Analyzer.


of

Assumption

Perfect Gyros and Accelerometers

Since this system is an angle measuring system which carries as

a reference an indication of the point of departure dependent on the accu-

racy of the gyros, any drift in the gyros which define the vertical plane
range and deflection. will give an equal angular error in

The third gyro,

SCI

CONFIDENT IAL
gLuo2)-Wl)(dIXP|]y
rravirrawrra sncnnaNid jo Noii33aio ONV 3NVTd i/VJNI NO NOIlVWm:>3V lNVlUlSU JO INlNOdWCD JO NCNlD3dK3 N33/K139 A3VW1D3VNJ WTTIONV 031VDIQNI

li
s

SNVWVUmHN
D8V JO S0NO335

za2 2 -zz Sap


I

iSs*

IIP 5*9
ZxujO
3UVJO S0N0D3S

a 1

9 UJUJ
< D Sz
o
2=i

SNViavamiw 5

(^^W =
SI XV

l(UJJD)-JOI]y

iNvid indNi ino wmnaN3d Nt 3DN33i3a HdV Ql 133d53b hi* wav 1N3WTI3 snonnoN3d JO NOI1V1N3I0 aVinONV Q31VD1QNI
I

4:

=lis

6
\s.

ZOzCO

o
2l
II

2.

UJ

ll

i i
&:

I
<

kOuj -uj 5: SOuJ


1

1
I i

f
BE-

<UJ

<

o5,^

a?

133

CONFIDENTIAL

E
<*

ex

S 9 i

2 ^

3 UJ
s
0E

UJ UJ oc
UJ
CI

S
= 2.
+
CT>

i
E

2 O a:

f
3
a*

UJ

o
CO oc or

8
cr

O u_

<
S 8
-

UJ

Si
3

3 UJ

sT

\n

^
cc

oc

*o
CJuj
r q;

Q
_

&

< H
Q:

UJ

a
jj>

UJ

? P
a

f
+

UO
U.UX

o
O
3"

<
P>
to

X
_)

li
X
I-

2!

io

O3
I

rj

13

UJ

z 2

DC

T
IX
1

1"

o
E a

o oc a: O o 2 o 2
uj

>

f
IS

UJ.

UJ dc

2 p

a?
"^UJ

UJ

O.QC o 2 3 UJ

8
z 3 8 u
z

I z * 5;

J34

CONHUtrtiiAL
which maintains the orientation
linear error.
of the track plane, will introduce a non-

The gyros have been assumed perfect primarily for simproblem.


If

plification of the

gyros of sufficient accuracy are found to

be impracticable, the necessary inertlal reference can be found from observation of the fixed stars.

Permanent errors

in the

zero indication of the pendulums also

introduce an error on the earth's surface equal in angle to the pendulum

error.
Single degree of freedom gyros and pendulous accelerometers of

types which

may prove

satisfactory are being developed in the Instrumen-

tation Laboratory at M.I.T.

These units are shown diagrammatically

in

Figs (A-6) and (A-7).


7.

Omission

of Constants of Integration

Wherever possible, throughout


have been assumed
to

this thesis, constants of integration

be zero. For example, in the range indication prob-

lem, the missile has been assumed to have no airspeed. This does not

change the time response or the size of the range or track error resulting

from wind disturbances, but

will have a large effect on the distance

travelled while the system is recovering

from a large disturbance.


to

At

all times, the

system has been assumed


wind accelerations
in the

be in equilibrium

when

hit with disturbing

time solutions of Chap-

ters in and IV.

135

tONHUtfiMrtl-

APPENDDC B
BIBLIOGRAPHY

1.

EHRENFRIED,

A. D.: "The Draper Notation

A Convenient System

of

Self -Defining Symbols." Instrumentation Laboratory,


of Aeronautical Engineering,

Department

Massachusetts Institute of Technology

October 1948.

2.

HUTZENLAUB,

J.

F.:

"Notebook dated July 21, 1947" Instrumentation

Laboratory, Massachusetts Institute of Technology.

3.

BOGOSIAN, A.

G.: "Notebook dated

March

18, 1947

Some General

Ifotes

on Physical State of Atmosphere" Instrumentation Laboratory,

Massachusetts

Institute of

Technology.

4.

GARDNER, M.

F. and

J.

L.

BARNES: "Transients

in

Linear Systems"

John Wiley

& Sons,

1942.

5.

Seminar

in

Automatic (Celestial and Inertial) Long -Range Guidance


the

Systems conducted at
the Scientific
to

Massachusetts Institute of Technology by


1

Advisory Board from

to 3

February, 1949 (Report

be published).
-

6.

"An Automatic Navigation System

Project Pebe"

USAF

Contract

W33 -03 8ac- 13969.

(Report to be published by Instrumentation

Laboratory, Massachusetts Institute of Technology).

7.

ROUTH, E. JJ "On

the Stability of a

Given State of Motion," Adams

Prize Essay, 1887.

8.

DEN HARTOG,
1934.

J.

P.:

"Mechanical Vibrations" McGraw-Hill Book Co.

136

CONFIDti

0.

BROWN,

G. 8. and D. P.

CAMPBELL:

"Principles of Servomechanisms"

John Wiley

Sons, 1948.

137

CONHUtNllAL

APPENDIX C
GLOSSARY

The system
because
1.

of notation

adopted for use in this thesis has been selected

it fulfills

the following desirable objectives:

It is It is

easily learned

2.

adaptable to a wide range of situations


built up

3.

It is

almost exclusively

of

characters found on the

keyboard of a standard American typewriter


4.

Any one

of the

compound symbols

of the

system

is readily

interpreted without recourse to an extensive glossary

few simple examples suffice

to explain the

operation of the notation

system. These examples are shown in table C~3. The short table of key

symbols given at the end

of this explanation will then furnish sufficient

information to enable any compund symbol to be correctly interpreted, and


to provide- the

necessary tools for the generation

of

new symbols.

A
table.

representative list of the primary symbols is given in the following

138

CONFIDtN'i lAL

Acceleration

A
(CT)

Angle
Characteristic time Characteristic time
-

[(CT)PR]
(C)

period ratio

Correction

(DR)
(Dir)

Dampting ratio
Direction

(DC)
(FR)

Dynamic Correction
Frequency ratio
Gravity

g
(PF)

Performance function
Radius
Sensitivity

R
S
(Sg)
t

Signal

Time
Velocity
Vertical direction

V
(Vert)

Frequency

PRIMARY SYMBOLS

TABLE C-l

139

CONFIDENTIAL

(aero)
(air)

Aerodynamic
Air mass

P
(pend)
(pu)

Performance operator

(d/dt)

Pendulum
Pendulous unit

(app)

Apparent value
Auxiliary

(aux)

r
(ref)

Range
Reference
Resultant

(CD)

Control direction
Controlled

(cm)
(dep)
(dest)

member

(res)
t

Departure
Destination

Time
Track Control;
trol plane
in track

(tc)

con-

E
(hor)
(ind)

Earth
(true)

True value
Vertical

Horizontal
(Vert)

Indicated value

X
IR
(kin)

Axis through nose and e.g.


of missile

Inertial reaction

Kinematical
Longitudinal plane; in range direction

Y
Z

Axis through right wing and


e.g. of missile

(long)

Axis through
lar to

e.g.

perpendicu

M
NE

XY

plane, directed

Missile

downward

Earth natural

MODIFYING SYMBOLS (SUBSCRIPTS)

TABLE C-2

V*0

CONFIULNiiAL

Symbol

Definition

V [E
S

(cm)]

Velocity of the controlled respect to the earth

member

with

[C(CD)][AA]

Sensitivity for correction of the control direction with angle input and rate of change of angle output

(C)A (Vert)(tc)

Correction to the angle of the vertical in the track control plane


Natural earth frequency equal to
g

W NE

IR

*E
A Z(cm)
(DC)
(ind)(tc)

Angle about the Z axis of the controlled

member
Indicated dynamic correction of track control system
Indicated dynamic correction of longitudinal system (i.e. range indication

^(indjflong)

system)

A r(ind)

Indicated range angle

Typical Examples Showing

How Symbols are Compounded from tary Forms

the

Elemen-

TABLE C-3

141

F L

2
I

60
i

550-DUP BINDERY
B
I

AP

3 6

IIDERY

Thesis E$65

Erb

^300G

A theoretical study of

automatic inertial navigation.


FE
2

60

-* -C
6

FEB

61

*SB3Tnv

S5Q-DUP

Thesis E565

43006
Erb A theoretical study of automatic inertial navigation.